app.py

import streamlit as st
import json
import ee
import os
import pandas as pd
import geopandas as gpd
from datetime import datetime
import leafmap.foliumap as leafmap
import time
import re

# Set up the page layout
st.set_page_config(layout="wide")

# Custom button styling
m = st.markdown(
    """
    <style>
    div.stButton > button:first-child {
        background-color: #006400;
        color:#ffffff;
    }
    </style>""",
    unsafe_allow_html=True,
)

# Logo
st.write(
    f"""
    <div style="display: flex; justify-content: space-between; align-items: center;">
        <img src="https://huggingface.co/spaces/YashMK89/GEE_Calculator/resolve/main/ISRO_Logo.png"  style="width: 20%; margin-right: auto;">
        <img src="https://huggingface.co/spaces/YashMK89/GEE_Calculator/resolve/main/SAC_Logo.png"  style="width: 20%; margin-left: auto;">
    </div>
    """,
    unsafe_allow_html=True,
)

# Authenticate and initialize Earth Engine
earthengine_credentials = os.environ.get("EE_Authentication")

# Initialize Earth Engine with secret credentials
os.makedirs(os.path.expanduser("~/.config/earthengine/"), exist_ok=True)
with open(os.path.expanduser("~/.config/earthengine/credentials"), "w") as f:
    f.write(earthengine_credentials)

ee.Initialize(project='ee-yashsacisro24')

# Load Sentinel dataset options from JSON file
with open("sentinel_datasets.json") as f:
    data = json.load(f)

# Display the title and dataset selection
st.title("Sentinel Dataset")

# Select dataset category and subcategory (case-insensitive selection)
main_selection = st.selectbox("Select Sentinel Dataset Category", list(data.keys()))

if main_selection:
    sub_options = data[main_selection]["sub_options"]
    sub_selection = st.selectbox("Select Specific Dataset ID", list(sub_options.keys()))

# Earth Engine Index Calculator Section
st.header("Earth Engine Index Calculator")

# Choose Index or Custom Formula (case-insensitive)
index_choice = st.selectbox("Select an Index or Enter Custom Formula", ['NDVI', 'NDWI', 'Average NO₂', 'Custom Formula'])

# Initialize custom_formula variable
custom_formula = ""

# Display corresponding formula based on the index selected (case-insensitive)
if index_choice.lower() == 'ndvi':
    st.write("Formula for NDVI: NDVI = (B8 - B4) / (B8 + B4)")
elif index_choice.lower() == 'ndwi':
    st.write("Formula for NDWI: NDWI = (B3 - B8) / (B3 + B8)")
elif index_choice.lower() == 'average no₂':
    st.write("Formula for Average NO₂: Average NO₂ = Mean(NO2 band)")
elif index_choice.lower() == 'custom formula':
    custom_formula = st.text_input("Enter Custom Formula (e.g., 'B5 - B4 / B5 + B4')")
    st.write(f"Custom Formula: {custom_formula}")  # Display the custom formula after the user inputs it

# Function to check if the polygon geometry is valid and convert it to the correct format
def convert_to_ee_geometry(geometry):
    # Ensure the polygon geometry is in the right format
    if geometry.is_valid:
        # Convert the geometry to GeoJSON format
        geojson = geometry.__geo_interface__
        # Convert to Earth Engine geometry
        return ee.Geometry(geojson)
    else:
        raise ValueError("Invalid geometry: The polygon geometry is not valid.")

# Function to read points from CSV
def read_csv(file_path):
    df = pd.read_csv(file_path)
    return df

# Function to read points from GeoJSON
def read_geojson(file_path):
    gdf = gpd.read_file(file_path)
    return gdf

# Function to read points from KML
def read_kml(file_path):
    gdf = gpd.read_file(file_path, driver='KML')
    return gdf

# Ask user whether they want to process 'Point' or 'Polygon' data (case-insensitive)
shape_type = st.selectbox("Do you want to process 'Point' or 'Polygon' data?", ["Point", "Polygon"])

# Ask user to upload a file based on shape type (case-insensitive)
file_upload = st.file_uploader(f"Upload your {shape_type} data (CSV, GeoJSON, KML)", type=["csv", "geojson", "kml"])

# Date Input for Start and End Dates
start_date = st.date_input("Start Date", value=pd.to_datetime('2020-01-01'))
end_date = st.date_input("End Date", value=pd.to_datetime('2020-12-31'))

# Convert start_date and end_date to string format for Earth Engine
start_date_str = start_date.strftime('%Y-%m-%d')
end_date_str = end_date.strftime('%Y-%m-%d')

# Initialize session state for storing results if not already done
if 'results' not in st.session_state:
    st.session_state.results = []
if 'last_params' not in st.session_state:
    st.session_state.last_params = {}
if 'map_data' not in st.session_state:
    st.session_state.map_data = None  # Initialize map_data

# Function to check if parameters have changed
def parameters_changed():
    return (
        st.session_state.last_params.get('main_selection') != main_selection or
        st.session_state.last_params.get('sub_selection') != sub_selection or
        st.session_state.last_params.get('index_choice') != index_choice or
        st.session_state.last_params.get('start_date_str') != start_date_str or
        st.session_state.last_params.get('end_date_str') != end_date_str
    )

# If parameters have changed, reset the results
if parameters_changed():
    st.session_state.results = []  # Clear the previous results
    # Update the last parameters to the current ones
    st.session_state.last_params = {
        'main_selection': main_selection,
        'sub_selection': sub_selection,
        'index_choice': index_choice,
        'start_date_str': start_date_str,
        'end_date_str': end_date_str
    }

# Function to perform index calculations
def calculate_ndvi(image, geometry):
    ndvi = image.normalizedDifference(['B8', 'B4']).rename('NDVI')
    result = ndvi.reduceRegion(
        reducer=ee.Reducer.mean(),
        geometry=geometry,
        scale=30
    )
    return result.get('NDVI')

def calculate_ndwi(image, geometry):
    ndwi = image.normalizedDifference(['B3', 'B8']).rename('NDWI')
    result = ndwi.reduceRegion(
        reducer=ee.Reducer.mean(),
        geometry=geometry,
        scale=30
    )
    return result.get('NDWI')

def calculate_avg_no2_sentinel5p(image, geometry):
    no2 = image.select('NO2').reduceRegion(
        reducer=ee.Reducer.mean(),
        geometry=geometry,
        scale=1000
    ).get('NO2')
    return no2

def calculate_custom_formula(image, geometry, formula):
    result = image.expression(formula).rename('Custom Index').reduceRegion(
        reducer=ee.Reducer.mean(),
        geometry=geometry,
        scale=30
    )
    return result.get('Custom Index')

# Process each point or polygon
if file_upload:
    locations_df = None  # Initialize locations_df to None
    polygons_df = None   # Initialize polygons_df to None

    file_extension = os.path.splitext(file_upload.name)[1].lower()  # Convert extension to lowercase

    # Read file based on shape type (case-insensitive)
    if shape_type.lower() == 'point':
        if file_extension == '.csv':
            locations_df = read_csv(file_upload)
        elif file_extension == '.geojson':
            locations_df = read_geojson(file_upload)
        elif file_extension == '.kml':
            locations_df = read_kml(file_upload)
        else:
            st.error("Unsupported file type. Please upload a CSV, GeoJSON, or KML file for points.")
    elif shape_type.lower() == 'polygon':
        if file_extension == '.geojson':
            polygons_df = read_geojson(file_upload)
        elif file_extension == '.kml':
            polygons_df = read_kml(file_upload)
        else:
            st.error("Unsupported file type. Please upload a GeoJSON or KML file for polygons.")
    
    # Check if locations_df is populated for points
    if locations_df is not None:
        # Display a preview of the points data
        st.write("Preview of the uploaded points data:")
        st.dataframe(locations_df.head())

        # Create a LeafMap object to display the points
        m = leafmap.Map(center=[locations_df['latitude'].mean(), locations_df['longitude'].mean()], zoom=10)

        # Add points to the map using a loop
        for _, row in locations_df.iterrows():
            latitude = row['latitude']
            longitude = row['longitude']
            
            # Check if latitude or longitude are NaN and skip if they are
            if pd.isna(latitude) or pd.isna(longitude):
                continue  # Skip this row and move to the next one
            
            m.add_marker(location=[latitude, longitude], popup=row.get('name', 'No Name'))

        # Display map
        st.write("Map of Uploaded Points:")
        m.to_streamlit()

        # Store the map in session_state
        st.session_state.map_data = m

        # Process each point for index calculation
        for idx, row in locations_df.iterrows():
            latitude = row['latitude']
            longitude = row['longitude']
            location_name = row.get('name', f"Point_{idx}")

            # Skip processing if latitude or longitude is NaN
            if pd.isna(latitude) or pd.isna(longitude):
                continue  # Skip this row and move to the next one

            # Define the region of interest (ROI)
            roi = ee.Geometry.Point([longitude, latitude])

            # Load Sentinel-2 image collection
            collection = ee.ImageCollection(sub_options[sub_selection]) \
                .filterDate(ee.Date(start_date_str), ee.Date(end_date_str)) \
                .filterBounds(roi)

            # Check if the collection has images for the selected date range
            image_count = collection.size().getInfo()
            if image_count == 0:
                st.warning(f"No images found for {location_name}.")
            else:
                st.write(f"Found {image_count} images for {location_name}.")
                image = collection.first()

                # Perform the calculation based on user selection
                result = None
                if index_choice.lower() == 'ndvi':
                    result = calculate_ndvi(image, roi)
                elif index_choice.lower() == 'ndwi':
                    result = calculate_ndwi(image, roi)
                elif index_choice.lower() == 'average no₂':
                    if 'NO2' in image.bandNames().getInfo():
                        result = calculate_avg_no2_sentinel5p(image, roi)
                    else:
                        st.warning(f"No NO2 band found for {location_name}. Please use Sentinel-5P for NO₂ data.")
                elif index_choice.lower() == 'custom formula' and custom_formula:
                    result = calculate_custom_formula(image, roi, custom_formula)

                if result is not None:
                    # Only store the numeric value (not the dictionary structure)
                    calculated_value = result.getInfo()  # Get the numeric value

                    # Store the result in session state
                    st.session_state.results.append({
                        'Location Name': location_name,
                        'Latitude': latitude,
                        'Longitude': longitude,
                        'Calculated Value': calculated_value
                    })

    # Check if polygons_df is populated for polygons
    if polygons_df is not None:
        # Display a preview of the polygons data
        st.write("Preview of the uploaded polygons data:")
        st.dataframe(polygons_df.head())

        # Create a LeafMap object to display the polygons
        m = leafmap.Map(center=[polygons_df.geometry.centroid.y.mean(), polygons_df.geometry.centroid.x.mean()], zoom=10)

        # Add polygons to the map
        for _, row in polygons_df.iterrows():
            polygon = row['geometry']
            if polygon.is_valid:  # Check if the geometry is valid
                # Create a GeoDataFrame with the single row
                gdf = gpd.GeoDataFrame([row], geometry=[polygon], crs=polygons_df.crs)
                
                # Add the valid GeoDataFrame to the map
                m.add_gdf(gdf=gdf, layer_name=row.get('name', 'Unnamed Polygon'))

        # Display map
        st.write("Map of Uploaded Polygons:")
        m.to_streamlit()

        # Store the map in session_state
        st.session_state.map_data = m

        # Process each polygon for index calculation
        for idx, row in polygons_df.iterrows():
            polygon = row['geometry']
            location_name = row.get('name', f"Polygon_{idx}")

            # Define the region of interest (ROI)
            try:
                roi = convert_to_ee_geometry(polygon)
            except ValueError as e:
                st.error(str(e))
                continue  # Skip this polygon if geometry is invalid

            # Load Sentinel-2 image collection
            collection = ee.ImageCollection(sub_options[sub_selection]) \
                .filterDate(ee.Date(start_date_str), ee.Date(end_date_str)) \
                .filterBounds(roi)

            # Check if the collection has images for the selected date range
            image_count = collection.size().getInfo()
            if image_count == 0:
                st.warning(f"No images found for {location_name}.")
            else:
                st.write(f"Found {image_count} images for {location_name}.")
                image = collection.first()

                # Perform the calculation based on user selection
                result = None
                if index_choice.lower() == 'ndvi':
                    result = calculate_ndvi(image, roi)
                elif index_choice.lower() == 'ndwi':
                    result = calculate_ndwi(image, roi)
                elif index_choice.lower() == 'average no₂':
                    if 'NO2' in image.bandNames().getInfo():
                        result = calculate_avg_no2_sentinel5p(image, roi)
                    else:
                        st.warning(f"No NO2 band found for {location_name}. Please use Sentinel-5P for NO₂ data.")
                elif index_choice.lower() == 'custom formula' and custom_formula:
                    result = calculate_custom_formula(image, roi, custom_formula)

                if result is not None:
                    # Only store the numeric value (not the dictionary structure)
                    calculated_value = result.getInfo()  # Get the numeric value

                    # Store the result in session state
                    st.session_state.results.append({
                        'Location Name': location_name,
                        'Calculated Value': calculated_value
                    })

# After processing, show the results
if st.session_state.results:
    # Convert the results to a DataFrame for better visualization
    result_df = pd.DataFrame(st.session_state.results)

    # If the shape type is 'Point', include 'Latitude' and 'Longitude'
    if shape_type.lower() == 'point':
        # Show the results in a table format with Latitude and Longitude
        st.write("Processed Results Table (Points):")
        st.dataframe(result_df[['Location Name', 'Latitude', 'Longitude', 'Calculated Value']])
    else:
        # For polygons, we only show the Location Name and Calculated Value
        st.write("Processed Results Table (Polygons):")
        st.dataframe(result_df[['Location Name', 'Calculated Value']])

    # Generate the dynamic filename
    filename = f"{main_selection}_{sub_selection}_{start_date.strftime('%Y/%m/%d')}_{end_date.strftime('%Y/%m/%d')}_{shape_type}.csv"

    # Convert results to DataFrame for download
    st.download_button(
        label="Download results as CSV",
        data=result_df.to_csv(index=False).encode('utf-8'),
        file_name=filename,
        mime='text/csv'
    )