upload_test_docs.py

import os
import requests
import time

API_URL = "https://diminch-tungtung-chatbot.hf.space"
API_KEY = "super-secret-internal-key"

headers = {
    "x-api-key": API_KEY
}

# 1. Chờ Space khởi động xong (Build and run)
print("Checking HF Space status...")
for i in range(30):
    try:
        r = requests.get(API_URL + "/")
        print(f"HF Space response status: {r.status_code}")
        if r.status_code in [200, 404]:
            print("HF Space is active!")
            break
    except Exception as e:
        print(f"Waiting for HF Space to start... ({e})")
    time.sleep(10)

# 2. Upload các tài liệu nghiệp vụ (business)
docs_dir = "test_documents"
if not os.path.exists(docs_dir):
    print(f"Directory {docs_dir} not found!")
    exit(1)

files = os.listdir(docs_dir)
for filename in files:
    filepath = os.path.join(docs_dir, filename)
    if not os.path.isfile(filepath):
        continue
    
    # Xác định category dựa trên tiền tố của tên file
    if filename.startswith("business"):
        category = "business"
    elif filename.startswith("learning"):
        category = "learning"
    else:
        continue
        
    print(f"\nUploading {filename} as category: {category}...")
    with open(filepath, "rb") as f:
        file_data = {"file": (filename, f, "text/plain")}
        params = {"doc_category": category, "uploaded_by": "admin"}
        try:
            r = requests.post(f"{API_URL}/upload", headers=headers, files=file_data, params=params)
            print(f"Status: {r.status_code}")
            try:
                print(r.json())
            except Exception:
                print(r.content.decode("utf-8", errors="ignore").encode("ascii", "replace").decode("ascii"))
        except Exception as e:
            print(f"Error uploading {filename}: {e}")

print("\nDone uploading all documents.")