Upload folder using huggingface_hub

app.py

@@ -1755,25 +1755,44 @@ def _uniprot_card(pmid):
     ], className="mb-3")
 
 
+_HF_REPO = "richiam/ProtoPure"
+_HF_APP_ROOT = "/app"
+
+
+def _resolve_xet_file(local_path):
+    """If local_path is an HF Xet pointer, download actual content and overwrite it."""
+    try:
+        with open(local_path, "rb") as f:
+            magic = f.read(2)
+        if magic == b"\x1f\x8b":
+            return  # already proper gzip
+        rel = os.path.relpath(local_path, _HF_APP_ROOT)
+        print(f"[xet] Downloading {rel} from HF Hub...", flush=True)
+        from huggingface_hub import hf_hub_download
+        import shutil
+        actual = hf_hub_download(repo_id=_HF_REPO, filename=rel, repo_type="space")
+        shutil.copy2(actual, local_path)
+        print(f"[xet] Cached {rel}", flush=True)
+    except Exception as e:
+        print(f"[xet] Failed to resolve {local_path}: {e}", flush=True)
+
+
 def _load_cluster_csv(model, min_val, threshold):
     """Return the ALL_FIELDS dataframe or None. Supports .csv.gz and .csv."""
     base = os.path.join(GRID_BASE, f"model={model}", f"min={min_val}",
                         f"t={threshold}_ALL_FIELDS")
-    print(f"[debug] _load_cluster_csv: GRID_BASE={GRID_BASE}, model={model}, min={min_val}, t={threshold}", flush=True)
     for ext in (".csv.gz", ".csv"):
         path = base + ext
-        exists = os.path.isfile(path)
-        print(f"[debug]   trying {path} -> exists={exists}", flush=True)
-        if not exists:
+        if not os.path.isfile(path):
             continue
+        if ext == ".csv.gz":
+            _resolve_xet_file(path)
         try:
             return pd.read_csv(path, compression="gzip" if ext == ".csv.gz" else None)
-        except Exception as e:
-            print(f"[debug]   read error ({e}), trying plain csv")
+        except Exception:
             try:
-                return pd.read_csv(path)
-            except Exception as e2:
-                print(f"[debug]   plain csv also failed: {e2}")
+                return pd.read_csv(path, compression=None)
+            except Exception:
                 continue
     return None
 
@@ -1962,11 +1981,9 @@ def _distribution_figure(cdf, field_name, top_n=20):
     Input("cluster-group-filter", "value"),
 )
 def update_cluster_plot(model, min_val, threshold, field, plot_type, top_n, groups):
-    print(f"[debug] update_cluster_plot called: model={model}, min={min_val}, t={threshold}, field={field}, plot_type={plot_type}", flush=True)
     empty_fig = go.Figure()
     empty_fig.update_layout(paper_bgcolor="#ffffff", plot_bgcolor="#f9fafc")
     if not all([model, min_val, threshold, field, plot_type]):
-        print(f"[debug] early return: missing values", flush=True)
         return empty_fig, "Select all parameters above."
 
     cdf = _load_cluster_csv(model, min_val, threshold)

requirements.txt

@@ -3,3 +3,4 @@ plotly==6.6.0
 dash-bootstrap-components==2.0.4
 pandas==3.0.2
 flask==3.1.3
+huggingface_hub>=0.23.0