Feature/UI

.streamlit/config.toml

@@ -0,0 +1,6 @@
+[theme]
+base = "light"
+primaryColor = "#2F6B9A"
+backgroundColor = "#FFFFFF"
+secondaryBackgroundColor = "#F8F8F8"
+textColor = "#000000"

Makefile

@@ -39,6 +39,9 @@ local_etl:
 	@USE_LOCAL=true poetry run python -m src.main.__main__
 	@echo "Hazard App run successfully"
 
+app:
+	@echo "Running Streamlit app..."
+	@PYTHONPATH=. poetry run streamlit run src/main/app.py
 
 help:
 	@echo "Available make targets:"

README.md

@@ -3,12 +3,27 @@
 A CLI tool to process hazard data (flood, earthquake,
 landslide, deforestation, cyclone, coastal erosion) and export results to AWS S3.
 
+## Poetry Usage
+
+Install Poetry (if not already installed):
+
+```bash
+curl -sSL https://install.python-poetry.org | python3 -
+```
+
+Project installation:
+
+```bash
+poetry --version
+```
+
 ## Features
 
 - Prepare exposure data from various sources
 - Process multiple hazards with xarray and geopandas
 - Export processed datasets to S3 using the VSI interface
 - Configurable via `src/utils/constants.py`
+- Interactive web app UI via Streamlit
 
 ## Requirements
 
@@ -33,6 +48,7 @@ make aws_etl   # Run the hazard processing pipeline
 make local_etl # Run the hazard processing pipeline locally without S3
 make test      # Run unit tests
 make lint      # Run lint checks
+make app      # Run the interactive Streamlit application
 ```
 
 ## Configuration

docs/app.md

@@ -0,0 +1,25 @@
+# Interactive Web App (app.py)
+
+Minimal guide for the Streamlit interface.
+
+## Running the App
+
+• `make app`
+• `PYTHONPATH=. poetry run streamlit run src/main/app.py`
+
+## Sidebar Controls
+
+• **Use Local ETL (no S3)**: save results locally
+• **Run Full Pipeline**: execute entire data workflow
+• **Select Hazard**: choose one of `flood`, `earthquake`, `landslide`,
+`deforestation`, `cyclone`, `coastal_erosion`
+• **Run [Hazard] Analysis**: process selected hazard and display results
+
+## Tabs
+
+• **Analysis**: display and download processed data as CSV
+• **Visualisation**: view previously saved results
+
+## File Location
+
+`src/main/app.py`

pyproject.toml

@@ -16,6 +16,9 @@ xarray = "^2025.3.1"
 pandas = "^2.2.3"
 rioxarray = "^0.18.2"
 rasterstats = "^0.20.0"
+streamlit = "^1.45.0"
+plotly = "^6.0.1"
+fsspec = "^2025.3.2"
 
 
 

src/main/app.py

@@ -0,0 +1,139 @@
+import importlib
+import os
+from datetime import datetime
+from pathlib import Path
+
+import geopandas as gpd
+import pandas as pd
+import rioxarray as rxr
+import streamlit as st
+from dotenv import load_dotenv
+
+import src.main.hazards as hazards
+from src.main.__main__ import main as run_full_pipeline
+from src.main.hazards import coastal_erosion, process_cyclone, process_deforestation
+from src.utils import constants
+from src.utils.utils import compute_hazard_mask, compute_population_exposure
+
+load_dotenv(dotenv_path=os.path.expanduser("~/.hazard_tool_rc"))
+
+
+def _lookup_opened_data(hazard_name: str):
+    """
+    Load the admin boundaries, population raster,
+    and preprocessed hazard exposure raster.
+    Returns: admin_gdf, pop_raster, exposure_raster
+    """
+    admin_gdf = gpd.read_file(constants.ADMIN_VECTOR_PATH)
+    pop_raster = rxr.open_rasterio(constants.POPULATION_RASTER_PATH, masked=True)
+    hazard_raster = rxr.open_rasterio(
+        constants.HAZARD_RASTER_PATH[hazard_name], masked=True
+    )
+    threshold = constants.HAZARD_THRESHOLD[hazard_name]
+
+    hazard_mask = compute_hazard_mask(hazard_raster, pop_raster, threshold)
+    exposure_raster = compute_population_exposure(hazard_mask, pop_raster)
+
+    return admin_gdf, pop_raster, exposure_raster
+
+
+st.set_page_config(page_title="Hazard Processing Tool", layout="wide")
+st.image("img/MA-logo.png", width=300)
+st.title("Hazard Processing Tool")
+st.markdown("Process hazard data and export results locally or to S3.")
+
+st.sidebar.header("Controls")
+use_local = st.sidebar.checkbox("Use Local ETL (no S3)", value=True)
+
+
+def _handle_run_full():
+    os.environ["USE_LOCAL"] = "true" if use_local else "false"
+    importlib.reload(constants)
+    import src.utils.s3 as s3mod
+
+    importlib.reload(s3mod)  # noqa: E702
+    import src.utils.utils as utilmod
+
+    importlib.reload(utilmod)  # noqa: E702
+    with st.spinner(f"Executing full pipeline ({'Local' if use_local else 'AWS'})..."):
+        run_full_pipeline()
+    st.sidebar.success("Pipeline complete!")
+
+
+if st.sidebar.button("Run Full Pipeline"):
+    _handle_run_full()
+
+hazard_options = [
+    "flood",
+    "earthquake",
+    "landslide",
+    "deforestation",
+    "cyclone",
+    "coastal_erosion",
+]
+hazard_choice = st.sidebar.selectbox("Select Hazard", hazard_options)
+tabs = st.tabs(["Analysis", "Visualisation"])
+
+with tabs[0]:
+    st.header("Hazard Analysis")
+
+    if st.button(f"Run {hazard_choice.capitalize()} Analysis"):
+        os.environ["USE_LOCAL"] = "true" if use_local else "false"
+        importlib.reload(constants)
+
+        with st.spinner(f"Processing {hazard_choice}..."):
+            if hazard_choice in ["flood", "earthquake", "landslide"]:
+                admin_gdf, pop_raster, exposure_raster = _lookup_opened_data(
+                    hazard_choice
+                )
+                df = hazards.process_hazards(admin_gdf, pop_raster, exposure_raster)
+            elif hazard_choice == "deforestation":
+                admin_gdf = gpd.read_file(constants.ADMIN_VECTOR_PATH)
+                df = process_deforestation(admin_gdf)
+            elif hazard_choice == "cyclone":
+                admin_gdf = gpd.read_file(constants.ADMIN_VECTOR_PATH)
+                df = process_cyclone(admin_gdf)
+            else:  # coastal_erosion
+                admin_gdf = gpd.read_file(constants.ADMIN_VECTOR_PATH)
+                df = coastal_erosion(admin_gdf)
+
+        st.success(f"{hazard_choice.capitalize()} analysis complete!")
+        st.session_state["result_df"] = df
+        st.dataframe(df, use_container_width=True)
+
+        # Save the result
+        output_path = constants.HAZARD_OUTPUT_PATH[hazard_choice]
+        if use_local:
+            Path(output_path).parent.mkdir(parents=True, exist_ok=True)
+            df.to_csv(output_path, index=False)
+        else:
+            # Still save locally even if tick was to S3
+            Path(output_path).parent.mkdir(parents=True, exist_ok=True)
+            df.to_csv(output_path, index=False)
+
+    if "result_df" in st.session_state:
+        csv_data = st.session_state["result_df"].to_csv(index=False).encode("utf-8")
+        ts = datetime.now().strftime("%Y%m%d_%H%M%S")
+        fname = f"{hazard_choice}_results_{ts}.csv"
+        st.download_button(
+            "Download Results as CSV", data=csv_data, file_name=fname, mime="text/csv"
+        )
+    else:
+        st.info("Click 'Run … Analysis' above to start.")
+
+with tabs[1]:
+    st.header("Visualisation")
+
+    if "result_df" in st.session_state:
+        df = st.session_state["result_df"]
+    else:
+        csv_path = Path(constants.HAZARD_OUTPUT_PATH[hazard_choice])
+        df = pd.read_csv(csv_path) if csv_path.exists() else None
+
+    if df is not None:
+        st.dataframe(df, use_container_width=True)
+    else:
+        st.info("Run an analysis (or the full pipeline) to visualise results.")
+
+st.markdown("---")
+st.markdown("Built on the hazard-processing-tool AWS pipeline.")