import streamlit as st
import pickle
import numpy as np
import pandas as pd
from sklearn.preprocessing import LabelEncoder
import re

# Load Data
df = pd.read_csv('Crop_Yield.csv')
cropOptions = list(df['Crop'].unique())
model_path = 'model.pkl'

css = """
<style>
    .stApp  {
        background-image: url("https://images.pexels.com/photos/265216/pexels-photo-265216.jpeg?auto=compress&cs=tinysrgb&w=1260&h=750&dpr=2");
        background-position: center;
        background-repeat: no-repeat;
        background-attachment: fixed;
        margin: 0;
        padding: 0;
    }
    .stForm{
        background-color: black;
    }
    .stButton > button {
        background-color: white;
        color: black;
    }
</style>
"""

# Inject custom CSS
st.markdown(css, unsafe_allow_html=True)

# Load Model
with open(model_path, 'rb') as file:
    model = pickle.load(file)

# Initialize Label Encoder and encode columns
label_encoder_crop = LabelEncoder()

df['Crop_encoded'] = label_encoder_crop.fit_transform(df['Crop'])

# Create mappings for Area and Item
crop_mapping = dict(zip(label_encoder_crop.classes_, range(len(label_encoder_crop.classes_))))

# Create Form
with st.form(key="my_form"):
    st.markdown("<h1 style='text-align: center; background-color: #f4edcd;color:black'>Crop Yield Prediction</h1>", unsafe_allow_html=True)
    
    crop = st.selectbox("Choose a Crop:", options=cropOptions)    
    area = st.text_input("Enter the Area in numbers  (in hectares):")
    area_error = "" if re.match(r"^\d+(\.\d+)?$", area) or not area else "Invalid input for area. Enter a numeric value without commas or special characters."
    if area_error:
        st.markdown(f"<span style='color:red;'>{area_error}</span>", unsafe_allow_html=True)
    
    production = st.text_input(" Enter the Production in numbers (in metric tons):")
    production_error = "" if re.match(r"^\d+(\.\d+)?$", production) or not production else "Invalid input for production. Enter a numeric value without commas or special characters."
    if production_error:
        st.markdown(f"<span style='color:red;'>{production_error}</span>", unsafe_allow_html=True)
    
    rainfall = st.slider("Annual Rainfall (in mm)", min(df['Annual_Rainfall']), max(df['Annual_Rainfall']), value=min(df['Annual_Rainfall']))
    fertilizer = st.slider("Fertilizer (in kilograms).", min(df['Fertilizer']), max(df['Fertilizer']), value=min(df['Fertilizer']))
    pesticide = st.slider("Pesticide (in kilograms).", min(df['Pesticide']), max(df['Pesticide']), value=min(df['Pesticide']))
    
    submit_button = st.form_submit_button(label="Predict")

# Handle Form Submission
if submit_button:
    # Validate Inputs
    if area_error or production_error:
        st.error("Please fix the errors above before proceeding.")
    else:
        # Prepare Input Data
        encoded_crop = crop_mapping[crop]

        input_data = np.array([[pesticide, fertilizer, rainfall,float(production), float(area),  encoded_crop]])
        
        # Predict
        try:
            prediction = model.predict(input_data)
            st.markdown(
                f"""
                <div style="color: black; font-size: 18px; border: 1px solid darkgreen; border-radius: 5px; padding: 10px; background-color: #e6ffe6;">
                    <strong>Expected Yield is (production per unit area):</strong> {prediction[0]}
                </div>
                """,
                unsafe_allow_html=True
            )
        except Exception as e:
            st.error(f"Error in prediction: {e}")