Update app/configs.py

app/configs.py

@@ -1,73 +1,138 @@
 import mlflow
 import mlflow.tensorflow
-
 from pathlib import Path
 from typing import Dict, Any, List
+import os
 
 # ======================
-# MLflow Setup
+# MLflow Configuration
 # ======================
-
-MLFLOW_URI = "https://omarelrayes-mlflow-server.hf.space"
+MLFLOW_URI = os.getenv("MLFLOW_TRACKING_URI", "https://omarelrayes-mlflow-server.hf.space")
 mlflow.set_tracking_uri(MLFLOW_URI)
 
+# HuggingFace Hub Configuration for Artifacts
+# ⚠️ IMPORTANT: HF_TOKEN must be set in Space Settings (not hardcoded!)
+HF_TOKEN = os.getenv("HF_TOKEN")
+if not HF_TOKEN:
+    raise ValueError(
+        "HF_TOKEN environment variable is not set! "
+        "Please add it in HuggingFace Space Settings -> Variables and secrets"
+    )
+os.environ["HF_TOKEN"] = HF_TOKEN
+
 # ======================
-# Model URIs
+# Model URIs (UPDATE WITH YOUR NEW RUN IDs)
 # ======================
+# ⚠️ IMPORTANT: Replace these with the actual run IDs from registration
+CLASSIFICATION_RUN_ID = os.getenv("CLASSIFICATION_RUN_ID")
+SEGMENTATION_RUN_ID = os.getenv("SEGMENTATION_RUN_ID")
+
+if not CLASSIFICATION_RUN_ID:
+    raise ValueError(
+        "CLASSIFICATION_RUN_ID environment variable is not set! "
+        "Please add it in HuggingFace Space Settings -> Variables and secrets"
+    )
+
+if not SEGMENTATION_RUN_ID:
+    raise ValueError(
+        "SEGMENTATION_RUN_ID environment variable is not set! "
+        "Please add it in HuggingFace Space Settings -> Variables and secrets"
+    )
 
-CLASSIFICATION_MODEL_URI = "runs:/2ae2c35a75a14f619fc782a9034b4a16/model"
-SEGMENTATION_MODEL_URI = "runs:/0835d2274eb34091aa209ef1800f92f0/model"
+CLASSIFICATION_MODEL_URI = f"runs:/{CLASSIFICATION_RUN_ID}/model"
+SEGMENTATION_MODEL_URI = f"runs:/{SEGMENTATION_RUN_ID}/model"
 
 # ======================
 # Cached models
 # ======================
-
 _classification_model = None
 _segmentation_model = None
 
-
 # ======================
-# Classification
+# Classification Model Loader
 # ======================
-
 def get_classification_model():
     global _classification_model
 
     if _classification_model is None:
-        print("Loading Classification Model...")
-
-        _classification_model = mlflow.tensorflow.load_model(
-            CLASSIFICATION_MODEL_URI
-        )
-
-        print("Classification Model Loaded")
+        print("="*60)
+        print("🔄 Loading Classification Model via MLflow...")
+        print(f"🔗 URI: {CLASSIFICATION_MODEL_URI}")
+        print(f"🌐 Tracking URI: {MLFLOW_URI}")
+        print("="*60)
+
+        try:
+            _classification_model = mlflow.tensorflow.load_model(
+                CLASSIFICATION_MODEL_URI
+            )
+            print("✅ Classification Model Loaded Successfully!")
+        except Exception as e:
+            print(f"❌ Error loading classification model: {e}")
+            print("\n🔍 Debugging Info:")
+            print(f"   - Tracking URI: {mlflow.get_tracking_uri()}")
+            print(f"   - Model URI: {CLASSIFICATION_MODEL_URI}")
+            
+            # Try to list available runs
+            try:
+                from mlflow.tracking import MlflowClient
+                client = MlflowClient()
+                runs = client.search_runs(experiment_names=["skin_cancer_classification"])
+                print(f"   - Available runs: {[run.info.run_id for run in runs]}")
+                
+                if runs:
+                    artifacts = client.list_artifacts(runs[0].info.run_id)
+                    print(f"   - Artifacts in first run: {[a.path for a in artifacts]}")
+            except Exception as debug_err:
+                print(f"   - Debug error: {debug_err}")
+            
+            raise
 
     return _classification_model
 
-
 # ======================
-# Segmentation
+# Segmentation Model Loader
 # ======================
-
 def get_segmentation_model():
     global _segmentation_model
 
     if _segmentation_model is None:
-        print("Loading Segmentation Model...")
-
-        _segmentation_model = mlflow.tensorflow.load_model(
-            SEGMENTATION_MODEL_URI
-        )
-
-        print("Segmentation Model Loaded")
+        print("="*60)
+        print("🔄 Loading Segmentation Model via MLflow...")
+        print(f"🔗 URI: {SEGMENTATION_MODEL_URI}")
+        print(f"🌐 Tracking URI: {MLFLOW_URI}")
+        print("="*60)
+
+        try:
+            _segmentation_model = mlflow.tensorflow.load_model(
+                SEGMENTATION_MODEL_URI
+            )
+            print("✅ Segmentation Model Loaded Successfully!")
+        except Exception as e:
+            print(f"❌ Error loading segmentation model: {e}")
+            print("\n🔍 Debugging Info:")
+            print(f"   - Tracking URI: {mlflow.get_tracking_uri()}")
+            print(f"   - Model URI: {SEGMENTATION_MODEL_URI}")
+            
+            # Try to list available runs
+            try:
+                from mlflow.tracking import MlflowClient
+                client = MlflowClient()
+                runs = client.search_runs(experiment_names=["skin_cancer_segmentation"])
+                print(f"   - Available runs: {[run.info.run_id for run in runs]}")
+                
+                if runs:
+                    artifacts = client.list_artifacts(runs[0].info.run_id)
+                    print(f"   - Artifacts in first run: {[a.path for a in artifacts]}")
+            except Exception as debug_err:
+                print(f"   - Debug error: {debug_err}")
+            
+            raise
 
     return _segmentation_model
 
-
 # ======================
-# Classes
+# Model Classes
 # ======================
-
 model_classes = {
     0: "benign",
     1: "malignant"
@@ -75,14 +140,14 @@ model_classes = {
 
 request_history: List[Dict[str, Any]] = []
 
-
 # ======================
-# Storage
+# Storage Directories
 # ======================
-
 STORAGE_DIR = Path("storage")
 IMAGES_DIR = STORAGE_DIR / "images"
 SEGMENTS_DIR = STORAGE_DIR / "segments"
 
 for d in [STORAGE_DIR, IMAGES_DIR, SEGMENTS_DIR]:
-    d.mkdir(parents=True, exist_ok=True)
+    d.mkdir(parents=True, exist_ok=True)
+
+print("✅ Models module initialized successfully!")