Update app.py

app.py

@@ -1,85 +1,117 @@
 # ===============================
-# Derma Space: Dataset Prep & HF Upload
+# Derma Space: Dataset + Gradio Chatbot
 # ===============================
 
 import json
 import random
 import os
+import gradio as gr
 from datasets import load_dataset
 from huggingface_hub import HfApi, login, upload_file
 
 # ------------------------
 # 1️⃣ ورود با Secret
 # ------------------------
-hf_token = os.environ["HF_TOKEN"]  # توکن باید در Secrets Space اضافه شده باشد
+hf_token = os.environ["HF_TOKEN"]
 login(token=hf_token)
 
 # ------------------------
-# 2️⃣ بارگذاری دیتاست عمومی
+# 2️⃣ ساخت دیتاست ترکیبی
 # ------------------------
-print("Loading DailyDialog...")
-dd = load_dataset("daily_dialog")['train']
-dd_examples = [{"domain":"general","context":conv[i],"response":conv[i+1]}
-               for conv in dd['dialog'] for i in range(len(conv)-1)]
+def build_dataset():
+    print("Loading DailyDialog...")
+    dd = load_dataset("daily_dialog")['train']
+    dd_examples = [{"domain":"general","context":conv[i],"response":conv[i+1]}
+                   for conv in dd['dialog'] for i in range(len(conv)-1)]
 
-print("Loading Persona-Chat...")
-pc = load_dataset("persona_chat", "self_original")['train']
-pc_examples = [{"domain":"general","context":conv['utterances'][i]['text'],
-                "response":conv['utterances'][i+1]['text']}
-               for conv in pc for i in range(len(conv['utterances'])-1)]
+    print("Loading Persona-Chat...")
+    pc = load_dataset("persona_chat", "self_original")['train']
+    pc_examples = [{"domain":"general","context":conv['utterances'][i]['text'],
+                    "response":conv['utterances'][i+1]['text']}
+                   for conv in pc for i in range(len(conv['utterances'])-1)]
 
-general_examples = dd_examples + pc_examples
-random.shuffle(general_examples)
-general_examples = general_examples[:5000]  # نمونه‌گیری
+    general_examples = dd_examples + pc_examples
+    random.shuffle(general_examples)
+    general_examples = general_examples[:5000]
 
-# ------------------------
-# 3️⃣ دیتاست تخصصی درماتولوژی
-# ------------------------
-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("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("Loading MedQuAD...")
-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']}
-                 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]
+    print("Loading MedQuAD...")
+    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']}
+                     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]
 
-dermatology_examples = derma_examples + medquad_derma
-random.shuffle(dermatology_examples)
+    dermatology_examples = derma_examples + medquad_derma
+    random.shuffle(dermatology_examples)
 
-# ------------------------
-# 4️⃣ ترکیب نهایی
-# ------------------------
-all_examples = general_examples + dermatology_examples
-random.shuffle(all_examples)
+    all_examples = general_examples + dermatology_examples
+    random.shuffle(all_examples)
 
-# ------------------------
-# 5️⃣ ذخیره به JSONL
-# ------------------------
-output_file = "derma_chat_mix.jsonl"
-with open(output_file, 'w', encoding='utf-8') as f:
-    for ex in all_examples:
-        f.write(json.dumps(ex, ensure_ascii=False) + "\n")
+    # ذخیره به JSONL
+    output_file = "derma_chat_mix.jsonl"
+    with open(output_file, 'w', encoding='utf-8') as f:
+        for ex in all_examples:
+            f.write(json.dumps(ex, ensure_ascii=False) + "\n")
+    print(f"✅ Dataset saved locally as {output_file} ({len(all_examples)} examples)")
 
-print(f"✅ Dataset saved locally as {output_file} ({len(all_examples)} examples)")
+    # آپلود به HF
+    repo_id = "username/Derma"  # نام کاربری خودت + نام Space
+    api = HfApi()
+    api.create_repo(repo_id=repo_id, repo_type="dataset", exist_ok=True)
+    upload_file(
+        path_or_fileobj=output_file,
+        path_in_repo=output_file,
+        repo_id=repo_id,
+        repo_type="dataset",
+        commit_message="Initial upload of text-based chat dataset"
+    )
+    print(f"✅ Dataset uploaded: https://huggingface.co/datasets/{repo_id}")
 
 # ------------------------
-# 6️⃣ آپلود به Hugging Face Hub
+# 3️⃣ چت ساده با Gradio
 # ------------------------
-repo_id = "username/Derma"  # نام کاربری خودت + نام Space
-api = HfApi()
-api.create_repo(repo_id=repo_id, repo_type="dataset", exist_ok=True)
+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]
+    
+    # جستجوی پاسخ بر اساس کلمات مشترک
+    best_match = None
+    max_overlap = 0
+    for item in data:
+        overlap = len(set(user_input.lower().split()) & set(item['context'].lower().split()))
+        if overlap > max_overlap:
+            max_overlap = overlap
+            best_match = item['response']
+    
+    if best_match:
+        return best_match
+    else:
+        return "Sorry, I don't have a good answer for that. Try another question!"
 
-upload_file(
-    path_or_fileobj=output_file,
-    path_in_repo=output_file,
-    repo_id=repo_id,
-    repo_type="dataset",
-    commit_message="Initial upload of text-based chat dataset"
+# ------------------------
+# 4️⃣ راه‌اندازی Gradio
+# ------------------------
+iface = gr.Interface(
+    fn=simple_chat,
+    inputs=gr.Textbox(lines=2, placeholder="Ask about dermatology or chat casually..."),
+    outputs=gr.Textbox(label="Derma ChatBot"),
+    title="Derma ChatBot",
+    description="A simple English chatbot combining general conversation + dermatology QA."
 )
 
-print(f"✅ Dataset uploaded to Hugging Face Hub: https://huggingface.co/datasets/{repo_id}")
+# ------------------------
+# 5️⃣ اجرای دیتاست + رابط
+# ------------------------
+if __name__ == "__main__":
+    # فقط یک بار دیتاست بساز
+    if not os.path.exists("derma_chat_mix.jsonl"):
+        build_dataset()
+    
+    iface.launch()