Update app.py

app.py

@@ -18,7 +18,7 @@ if hf_token is None:
 login(token=hf_token)
 
 # ------------------------
-# 2️⃣ ساخت دیتاست ترکیبی
+# 2️⃣ ساخت دیتاست ترکیبی امن
 # ------------------------
 def build_dataset():
     print("Creating a small general dataset...")
@@ -27,15 +27,25 @@ def build_dataset():
         {"domain":"general", "context":"What's your name?", "response":"I'm Derma ChatBot."}
     ]
 
+    # ----- Dermatology QA (Mreeb)
     print("Loading Dermatology QA (Mreeb)...")
     derma = load_dataset("Mreeb/Dermatology-Question-Answer-Dataset-For-Fine-Tuning")['train']
-    derma_examples = [{"domain":"dermatology","context":item['question'],"response":item['answer']}
-                      for item in derma]
+    print("Columns in Mreeb dataset:", derma.column_names)
 
+    derma_examples = []
+    for item in derma:
+        q = item.get('question') or item.get('Question') or item.get('Q')
+        a = item.get('answer') or item.get('Answer') or item.get('A')
+        if q and a:
+            derma_examples.append({"domain":"dermatology","context":q,"response":a})
+
+    # ----- MedQuAD subset
     print("Loading MedQuAD subset...")
     medquad = load_dataset("pythonafroz/MedQuAD")['train']
     derma_keywords = ["skin", "eczema", "psoriasis", "dermatitis", "melanoma", "acne", "rash"]
-    medquad_derma = [{"domain":"dermatology","context":item['question'],"response":item['answer']}
+    medquad_derma = [{"domain":"dermatology",
+                      "context":item['question'],
+                      "response":item['answer']}
                      for item in medquad if any(k in item['question'].lower() for k in derma_keywords)]
     random.shuffle(medquad_derma)
     medquad_derma = medquad_derma[:500]
@@ -70,10 +80,9 @@ def build_dataset():
 # 3️⃣ چت ساده با Gradio
 # ------------------------
 def simple_chat(user_input):
-    # جستجو در دیتاست برای پاسخ نزدیک (ساده)
     with open("derma_chat_mix.jsonl", 'r', encoding='utf-8') as f:
         data = [json.loads(line) for line in f]
-    
+
     best_match = None
     max_overlap = 0
     for item in data:
@@ -81,7 +90,7 @@ def simple_chat(user_input):
         if overlap > max_overlap:
             max_overlap = overlap
             best_match = item['response']
-    
+
     if best_match:
         return best_match
     else: