Update README and app.py for energy file handling and visualization details

README.md

@@ -24,12 +24,15 @@ For the selected weather/building pair, the app fetches:
 - `geometry/<building_id>.npz`
 - `building/<building_id>.npz`
 - `weather/<weather_id>.npz`
-- `energy/<sample_id>.npz`
+- `energy/<energy_file from manifest.csv>`
 
-Then it renders four views:
+The energy file can live under nested subfolders such as `energy/00/00017__Anchorage.npz`.
+
+Then it renders five views:
 
 - Geometry polygons
 - Building graph
+- Graph/geometry overlap
 - Weather curves (selected time window)
 - Energy curves (selected time window)
 

app.py

@@ -6,6 +6,7 @@ import time
 from dataclasses import dataclass
 from functools import lru_cache
 from pathlib import Path
+from pathlib import PurePosixPath
 
 import gradio as gr
 from huggingface_hub import hf_hub_download
@@ -148,28 +149,40 @@ def _resolve_record(weather_id: str, building_id: str) -> SampleRecord:
     )
 
 
+def _repo_dataset_path(*parts: str) -> str:
+    clean_parts: list[str] = []
+    for part in parts:
+        text = (part or "").replace("\\", "/").strip("/")
+        if text:
+            clean_parts.extend(segment for segment in text.split("/") if segment and segment != ".")
+    return str(PurePosixPath(*clean_parts))
+
+
 def _download_modalities(record: SampleRecord) -> tuple[Path, Path, Path, Path]:
     building_key = record.sample_id.split("__", 1)[0]
+    energy_relpath = record.energy_file.replace("\\", "/").lstrip("/")
+    if energy_relpath.startswith("energy/"):
+        energy_relpath = energy_relpath[len("energy/") :]
 
     geometry_npz = hf_hub_download(
         repo_id=DATASET_REPO_ID,
         repo_type="dataset",
-        filename=f"geometry/{building_key}.npz",
+        filename=_repo_dataset_path("geometry", f"{building_key}.npz"),
     )
     graph_npz = hf_hub_download(
         repo_id=DATASET_REPO_ID,
         repo_type="dataset",
-        filename=f"building/{building_key}.npz",
+        filename=_repo_dataset_path("building", f"{building_key}.npz"),
     )
     weather_npz = hf_hub_download(
         repo_id=DATASET_REPO_ID,
         repo_type="dataset",
-        filename=f"weather/{record.weather_id}.npz",
+        filename=_repo_dataset_path("weather", f"{record.weather_id}.npz"),
     )
     energy_npz = hf_hub_download(
         repo_id=DATASET_REPO_ID,
         repo_type="dataset",
-        filename=f"energy/{record.energy_file}",
+        filename=_repo_dataset_path("energy", energy_relpath),
     )
 
     return Path(geometry_npz), Path(graph_npz), Path(weather_npz), Path(energy_npz)