utils.py

import pandas as pd
import numpy as np
import plotly.graph_objects as go
import plotly.express as px
import xarray as xr
import solara
from uuid import uuid4
from sklearn.neighbors import NearestNeighbors
from scipy.stats import linregress
import geopandas as gpd
from constants import REGION_COLUMN
from Datasets import DATASETS

def get_plotly_theme():
    return "plotly_dark" if solara.lab.theme.dark_effective else "plotly_white"

def load_dataset(dataset_config):
    dataset = xr.open_dataset(dataset_config["average_path"], decode_timedelta=False)
    variables = list(dataset.data_vars)
    float_vars = [var for var in variables if dataset[var].dtype in ["float32", "float64"]]
    
    data = dataset.to_dataframe().reset_index().dropna()
    data = data.rename(columns={"latitude": "lat", "longitude": "lon"})
    
    if len(data) >= 4:
        sample_indices = np.random.choice(len(data), size=4, replace=False)
        sample_coords = data[["lat", "lon"]].iloc[sample_indices].values
        coords = data[["lat", "lon"]].values
        
        knn = NearestNeighbors(n_neighbors=2).fit(coords)
        distances, _ = knn.kneighbors(sample_coords, n_neighbors=2)
        grid_spacing = distances[:, 1].mean()
    else:
        grid_spacing = 0.25
    
    dataset.close()
    return data, float_vars, grid_spacing

def calculate_map_zoom(lat_range, lon_range):
    range_max = max(lat_range, lon_range)
    return (1 if range_max > 100 else 
            2 if range_max > 30 else 
            4 if range_max > 15 else 
            5 if range_max > 5 else 
            7 if range_max > 1 else 
            9 if range_max > 0.1 else 10)

def create_heatmap(data, dataset_name, value_column, grid_size, color_scale, map_style, opacity):
    center = {"lat": data["lat"].mean(), "lon": data["lon"].mean()}
    zoom = calculate_map_zoom(data["lat"].max() - data["lat"].min(), 
                            data["lon"].max() - data["lon"].min())
    
    features = [
        {
            "type": "Feature", "id": str(uuid4()),
            "geometry": {
                "type": "Polygon",
                "coordinates": [[
                    [row["lon"] - grid_size/2, row["lat"] - grid_size/2],
                    [row["lon"] + grid_size/2, row["lat"] - grid_size/2],
                    [row["lon"] + grid_size/2, row["lat"] + grid_size/2],
                    [row["lon"] - grid_size/2, row["lat"] + grid_size/2],
                    [row["lon"] - grid_size/2, row["lat"] - grid_size/2]
                ]]
            },
            "properties": {"value": row[value_column], "lat": row["lat"], "lon": row["lon"]}
        }
        for _, row in data.iterrows()
    ]
    
    geojson = {"type": "FeatureCollection", "features": features}
    fig = go.Figure(
        go.Choroplethmap(
            geojson=geojson,
            locations=[f["id"] for f in features],
            z=[f["properties"]["value"] for f in features],
            colorscale=color_scale,
            marker_opacity=opacity,
            marker_line_width=0,
            colorbar=dict(title=dataset_name),
            customdata=np.stack(([f["properties"]["lat"] for f in features], 
                               [f["properties"]["lon"] for f in features]), axis=-1),
            hovertemplate="<b>Lat: %{customdata[0]:.3f}</b><br>Lon: %{customdata[1]:.3f}<br>Value: %{z}<extra></extra>"
        )
    )
    fig.update_layout(
        map=dict(style=map_style, center=center, zoom=zoom),
        height=900, 
        width=1250,
        title=f"Heatmap - {dataset_name}",
        template=get_plotly_theme()
    )
    return fig

def create_timeseries_plot(file_path, dataset_name, lat, lon, plotted_line_color, spatial_mean_data=None, show_spatial_mean=False, from_year=None, to_year=None, use_timeframe=False, region="India", show_trend_line=False, trend_line_color="red", spatial_mean_color="yellow"):
    try:
        dataset = xr.open_dataset(file_path, decode_timedelta=False)
        variable = list(dataset.data_vars)[0]
        
        lat_idx = np.abs((dataset.get("lat", dataset.get("latitude")).values - lat)).argmin()
        lon_idx = np.abs((dataset.get("lon", dataset.get("longitude")).values - lon)).argmin()
        
        timeseries = dataset[variable].isel(lat=lat_idx, lon=lon_idx).to_dataframe().reset_index()
        
        years = timeseries["year"].values
        if len(years) == 0:
            raise ValueError("No years found in the dataset")
        start_year = int(min(years))
        end_year = int(max(years))
        
        if use_timeframe and from_year is not None and to_year is not None:
            timeseries = timeseries[(timeseries["year"] >= from_year) & (timeseries["year"] <= to_year)]
            start_year = from_year
            end_year = to_year
        
        x = timeseries["year"].values
        y = timeseries[variable].values
        
        # Fit parabolic trend to actual data
        if show_trend_line and len(x) > 2:
            x_norm = x - x.mean()
            coeffs = np.polyfit(x_norm, y, deg=2)
            poly = np.poly1d(coeffs)
            trend_line = poly(x_norm)
            midpoint_idx = len(x) // 2
            slope = 2 * coeffs[0] * x_norm[midpoint_idx] + coeffs[1]
            _, _, r_value, p_value, _ = linregress(x, y)
            trend = "Increasing" if p_value < 0.05 and slope > 0 else "Decreasing" if p_value < 0.05 and slope < 0 else "No Significant Trend"
        else:
            trend_line = np.full_like(x, np.nan)
            slope = 0
            p_value = 1.0
            trend = "No Significant Trend"
        
        print(f"Creating timeseries plot: dataset_name={dataset_name}, start_year={start_year}, end_year={end_year}, lat={lat:.2f}, lon={lon:.2f}, region={region}")
        
        fig = px.line(
            timeseries, 
            x="year", 
            y=variable,
            title=f"{dataset_name}, {start_year}-{end_year} for Lat: {lat:.2f}, Lon: {lon:.2f}",
            labels={"year": "Year", variable: dataset_name},
            color_discrete_sequence=[plotted_line_color],
            render_mode="svg"
        )
        fig.data[0].name = dataset_name
        
        if show_trend_line and len(x) > 2:
            fig.add_scatter(
                x=x, 
                y=trend_line,
                mode="lines", 
                name=f"{dataset_name} Trend",
                line=dict(color=trend_line_color, dash="dash")
            )
        
        if show_spatial_mean and spatial_mean_data is not None:
            spatial_mean_df = spatial_mean_data.get("timeseries", pd.DataFrame())
            if not spatial_mean_df.empty:
                fig.add_scatter(
                    x=spatial_mean_df["year"],
                    y=spatial_mean_df[variable],
                    mode="lines",
                    name=f"{region} Spatial Mean",
                    line=dict(color=spatial_mean_color, width=2)
                )
        
        fig.add_annotation(
            x=timeseries["year"].max(),
            y=timeseries[variable].min(),
            text=f"Slope: {slope:.4f}<br>p-value: {p_value:.4f}<br>Trend: {trend}",
            showarrow=False, 
            xanchor="right", 
            yanchor="top",
            bgcolor="white" if not solara.lab.theme.dark_effective else "#333333",
            bordercolor="black", 
            borderpad=4, 
            opacity=0.8,
            font=dict(size=12, color="black" if not solara.lab.theme.dark_effective else "white")
        )
        
        fig.update_layout(
            xaxis_title="Year", 
            yaxis_title=dataset_name,
            height=550, 
            width=650, 
            showlegend=True,
            legend=dict(
                x=0.99,
                y=0.99,
                xanchor="right",
                yanchor="top",
                bgcolor="rgba(0, 0, 0, 0)"
            ),
            template=get_plotly_theme()
        )
        
        dataset.close()
        return fig
    
    except Exception as e:
        print(f"Error creating timeseries plot: {e}")
        dataset.close() if 'dataset' in locals() else None
        return go.Figure()

def get_region_bounds(geodataframe, region):
    geometry = geodataframe[geodataframe[REGION_COLUMN] == region].geometry.iloc[0]
    minx, miny, maxx, maxy = geometry.bounds
    return {"min_lon": minx, "max_lon": maxx, "min_lat": miny, "max_lat": maxy}

def filter_region_data(data, geodataframe, region):
    if not region or region == "India":
        return data
    
    try:
        points = gpd.GeoDataFrame(
            data, geometry=gpd.points_from_xy(data["lon"], data["lat"]), crs=geodataframe.crs
        )
        points.sindex
        
        region_geometry = geodataframe[geodataframe[REGION_COLUMN] == region].geometry.iloc[0]
        
        bbox = region_geometry.bounds
        candidates = points[points.geometry.bounds.minx >= bbox[0]]
        candidates = candidates[candidates.geometry.bounds.maxx <= bbox[2]]
        candidates = candidates[candidates.geometry.bounds.miny >= bbox[1]]
        candidates = candidates[candidates.geometry.bounds.maxy <= bbox[3]]
        
        filtered_points = candidates[candidates.geometry.within(region_geometry)]
        
        return filtered_points.drop(columns=["geometry"])
    
    except Exception as e:
        print(f"Error filtering region data: {e}")
        return pd.DataFrame()

def calculate_spatial_mean(dataset_config, geodataframe, region, from_year=None, to_year=None, use_timeframe=False):
    try:
        dataset = xr.open_dataset(dataset_config["timeseries_path"], decode_timedelta=False)
        variable = list(dataset.data_vars)[0]
        no_data_value = dataset_config["no_data"]

        data = dataset[variable].to_dataframe().reset_index()
        data = data.rename(columns={"latitude": "lat", "longitude": "lon"})

        data = data[data[variable] != no_data_value]

        if use_timeframe and from_year is not None and to_year is not None:
            data = data[(data["year"] >= from_year) & (data["year"] <= to_year)]

        if region != "India" and geodataframe is not None:
            data = filter_region_data(data, geodataframe, region)
        
        spatial_means = data.groupby("year")[variable].mean().reset_index()
        overall_mean = spatial_means[variable].mean()
        
        dataset.close()
        return {"overall_mean": overall_mean, "timeseries": spatial_means}
    
    except Exception as e:
        print(f"Error calculating spatial mean: {e}")
        return {"overall_mean": None, "timeseries": pd.DataFrame()}

def calculate_selected_spatial_mean(dataset_config, selected_coords, from_year=None, to_year=None, use_timeframe=False):
    try:
        if not selected_coords:
            return {"overall_mean": None, "timeseries": pd.DataFrame()}
        
        dataset = xr.open_dataset(dataset_config["timeseries_path"], decode_timedelta=False)
        variable = list(dataset.data_vars)[0]
        no_data_value = dataset_config["no_data"]

        data_list = []
        for lat, lon in selected_coords:
            lat_idx = np.abs((dataset.get("lat", dataset.get("latitude")).values - lat)).argmin()
            lon_idx = np.abs((dataset.get("lon", dataset.get("longitude")).values - lon)).argmin()
            point_data = dataset[variable].isel(lat=lat_idx, lon=lon_idx).to_dataframe().reset_index()
            data_list.append(point_data)
        
        data = pd.concat(data_list, ignore_index=True)
        data = data[data[variable] != no_data_value]

        if use_timeframe and from_year is not None and to_year is not None:
            data = data[(data["year"] >= from_year) & (data["year"] <= to_year)]

        spatial_means = data.groupby("year")[variable].mean().reset_index()
        overall_mean = spatial_means[variable].mean()
        
        dataset.close()
        return {"overall_mean": overall_mean, "timeseries": spatial_means}
    
    except Exception as e:
        print(f"Error calculating selected spatial mean: {e}")
        return {"overall_mean": None, "timeseries": pd.DataFrame()}

def create_selected_timeseries_plot(file_path, dataset_name, selected_coords, plotted_line_color, selected_spatial_mean_data=None, from_year=None, to_year=None, use_timeframe=False, show_trend_line=False, trend_line_color="red", spatial_mean_color="yellow"):
    try:
        if not selected_coords:
            return go.Figure()
        
        dataset = xr.open_dataset(file_path, decode_timedelta=False)
        variable = list(dataset.data_vars)[0]
        no_data_value = next(d for d in DATASETS if d["name"] == dataset_name)["no_data"]

        data_list = []
        for lat, lon in selected_coords:
            lat_idx = np.abs((dataset.get("lat", dataset.get("latitude")).values - lat)).argmin()
            lon_idx = np.abs((dataset.get("lon", dataset.get("longitude")).values - lon)).argmin()
            point_data = dataset[variable].isel(lat=lat_idx, lon=lon_idx).to_dataframe().reset_index()
            data_list.append(point_data)
        
        data = pd.concat(data_list, ignore_index=True)
        data = data[data[variable] != no_data_value]

        timeseries = data.groupby("year")[variable].mean().reset_index()

        years = timeseries["year"].values
        if len(years) == 0:
            raise ValueError("No years found in the dataset")
        start_year = int(min(years))
        end_year = int(max(years))

        if use_timeframe and from_year is not None and to_year is not None:
            timeseries = timeseries[(timeseries["year"] >= from_year) & (timeseries["year"] <= to_year)]
            start_year = from_year
            end_year = to_year

        x = timeseries["year"].values
        y = timeseries[variable].values
        
        # Fit parabolic trend to actual data
        if show_trend_line and len(x) > 2:
            x_norm = x - x.mean()
            coeffs = np.polyfit(x_norm, y, deg=2)
            poly = np.poly1d(coeffs)
            trend_line = poly(x_norm)
            midpoint_idx = len(x) // 2
            slope = 2 * coeffs[0] * x_norm[midpoint_idx] + coeffs[1]
            _, _, r_value, p_value, _ = linregress(x, y)
            trend = "Increasing" if p_value < 0.05 and slope > 0 else "Decreasing" if p_value < 0.05 and slope < 0 else "No Significant Trend"
        else:
            trend_line = np.full_like(x, np.nan)
            slope = 0
            p_value = 1.0
            trend = "No Significant Trend"
        
        fig = px.line(
            timeseries, 
            x="year", 
            y=variable,
            title=f"{dataset_name}, Spatial Mean of Selected Points, {start_year}-{end_year}",
            labels={"year": "Year", variable: dataset_name},
            color_discrete_sequence=[plotted_line_color]
        )
        fig.data[0].name = dataset_name
        
        if show_trend_line and len(x) > 2:
            fig.add_scatter(
                x=x, 
                y=trend_line,
                mode="lines", 
                name=f"{dataset_name} Trend",
                line=dict(color=trend_line_color, dash="dash")
            )
        
        if selected_spatial_mean_data is not None:
            spatial_mean_df = selected_spatial_mean_data.get("timeseries", pd.DataFrame())
            if not spatial_mean_df.empty:
                fig.add_scatter(
                    x=spatial_mean_df["year"],
                    y=spatial_mean_df[variable],
                    mode="lines",
                    name="Selected Points Spatial Mean",
                    line=dict(color=spatial_mean_color, width=2)
                )
        
        fig.add_annotation(
            x=timeseries["year"].max(),
            y=timeseries[variable].min(),
            text=f"Slope: {slope:.4f}<br>p-value: {p_value:.4f}<br>Trend: {trend}",
            showarrow=False, 
            xanchor="right", 
            yanchor="top",
            bgcolor="white" if not solara.lab.theme.dark_effective else "#333333",
            bordercolor="black", 
            borderpad=4, 
            opacity=0.8,
            font=dict(size=12, color="black" if not solara.lab.theme.dark_effective else "white")
        )
        
        fig.update_layout(
            xaxis_title="Year", 
            yaxis_title=f"Mean {dataset_name}",
            height=550, 
            width=650, 
            showlegend=True,
            legend=dict(
                x=0.99,
                y=0.99,
                xanchor="right",
                yanchor="top",
                bgcolor="rgba(0, 0, 0, 0)"
            ),
            template=get_plotly_theme()
        )
        
        dataset.close()
        return fig
    
    except Exception as e:
        print(f"Error creating selected timeseries plot: {e}")
        dataset.close() if 'dataset' in locals() else None
        return go.Figure()

def get_available_years(dataset_config):
    try:
        dataset = xr.open_dataset(dataset_config["timeseries_path"], decode_timedelta=False)
        years = dataset["year"].values.tolist()
        dataset.close()
        return sorted(years)
    except Exception as e:
        print(f"Error getting available years: {e}")
        return []