"""
Hugging Face Space 应用：在 HF Space 上运行 Weaviate 索引构建
默认使用 HF 免费 embedding（sentence-transformers），直接上传到 Weaviate Cloud
"""
import os
import gradio as gr
from pathlib import Path
import threading
import time

# 从环境变量读取配置（HF Space Secrets）
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY", "").strip()
WEAVIATE_URL = os.getenv("WEAVIATE_URL", "").strip()
WEAVIATE_API_KEY = os.getenv("WEAVIATE_API_KEY", "").strip()
WEAVIATE_COLLECTION = os.getenv("WEAVIATE_COLLECTION", "GenAICourses").strip()
EMBEDDING_PROVIDER = os.getenv("EMBEDDING_PROVIDER", "huggingface").strip().lower()

# 课程文档路径（需要上传到 HF Space）
SCRIPT_DIR = Path(__file__).resolve().parent
COURSES_DIR = SCRIPT_DIR / "GENAI COURSES"

# 全局状态
build_status = {"running": False, "progress": "", "error": None, "result": None}


def build_index_worker(clear_first: bool, progress_callback=None):
    """后台工作线程：构建索引"""
    global build_status
    
    try:
        build_status["running"] = True
        build_status["error"] = None
        build_status["progress"] = "开始构建索引..."
        
        # 检查配置
        if EMBEDDING_PROVIDER == "openai" and not OPENAI_API_KEY:
            raise RuntimeError("使用 OpenAI embedding 时请在 Settings → Secrets 中添加 OPENAI_API_KEY")
        if not WEAVIATE_URL or not WEAVIATE_API_KEY:
            raise RuntimeError("请在 HF Space Settings → Secrets 中添加 WEAVIATE_URL 和 WEAVIATE_API_KEY")
        
        # 检查课程目录
        if not COURSES_DIR.exists():
            raise FileNotFoundError(
                f"课程目录不存在：{COURSES_DIR}\n"
                "请将 GENAI COURSES 文件夹上传到 Space 的根目录"
            )
        
        # 导入依赖
        build_status["progress"] = "加载依赖库..."
        from llama_index.core import SimpleDirectoryReader, VectorStoreIndex, Settings
        from llama_index.core import StorageContext
        from llama_index.vector_stores.weaviate import WeaviateVectorStore
        import weaviate
        from weaviate.classes.init import Auth
        
        # 设置 embedding
        build_status["progress"] = "配置 embedding 模型..."
        if EMBEDDING_PROVIDER == "openai":
            from llama_index.embeddings.openai import OpenAIEmbedding
            Settings.embed_model = OpenAIEmbedding(
                model="text-embedding-3-small",
                api_key=OPENAI_API_KEY,
            )
        else:
            from llama_index.embeddings.huggingface import HuggingFaceEmbedding
            Settings.embed_model = HuggingFaceEmbedding(
                model_name="sentence-transformers/all-MiniLM-L6-v2"
            )
        
        # 连接 Weaviate
        build_status["progress"] = "连接 Weaviate Cloud..."
        url = WEAVIATE_URL
        if not url.startswith("http"):
            url = "https://" + url
        
        client = weaviate.connect_to_weaviate_cloud(
            cluster_url=url,
            auth_credentials=Auth.api_key(WEAVIATE_API_KEY),
        )
        
        if not client.is_ready():
            raise RuntimeError("Weaviate 连接失败")
        
        try:
            # 清空旧 collection（如果需要）
            if clear_first:
                build_status["progress"] = f"删除旧 collection: {WEAVIATE_COLLECTION}..."
                try:
                    if hasattr(client.collections, "delete"):
                        client.collections.delete(WEAVIATE_COLLECTION)
                        build_status["progress"] = "旧 collection 已删除"
                except Exception as e:
                    if "404" not in str(e) and "not found" not in str(e).lower():
                        build_status["progress"] = f"删除旧 collection 时警告: {e}"
            
            # 读取文档
            build_status["progress"] = f"读取课程目录: {COURSES_DIR}..."
            reader = SimpleDirectoryReader(
                input_dir=str(COURSES_DIR),
                recursive=True,
                required_exts=[".md", ".pdf", ".txt", ".py", ".ipynb", ".docx"],
            )
            documents = reader.load_data()
            build_status["progress"] = f"已加载 {len(documents)} 个文档块"
            
            # 创建 vector store
            build_status["progress"] = "创建 Weaviate vector store..."
            vector_store = WeaviateVectorStore(
                weaviate_client=client,
                index_name=WEAVIATE_COLLECTION,
            )
            storage_context = StorageContext.from_defaults(vector_store=vector_store)
            
            # 构建索引（这会自动进行 embedding 并上传）
            build_status["progress"] = f"正在 embedding 并上传到 Weaviate (collection={WEAVIATE_COLLECTION})...\n这可能需要几分钟时间，请耐心等待..."
            index = VectorStoreIndex.from_documents(
                documents,
                storage_context=storage_context,
            )
            
            # 等待 batch 提交完成
            time.sleep(3)
            
            # 验证
            build_status["progress"] = "验证索引..."
            coll = client.collections.get(WEAVIATE_COLLECTION)
            agg = coll.aggregate.over_all(total_count=True)
            n = agg.total_count
            
            build_status["result"] = f"✅ 索引构建成功！\n当前 object count = {n}"
            build_status["progress"] = build_status["result"]
            
        finally:
            client.close()
            
    except Exception as e:
        build_status["error"] = str(e)
        build_status["progress"] = f"❌ 错误: {str(e)}"
    finally:
        build_status["running"] = False


def start_build(clear_first: bool):
    """启动索引构建"""
    global build_status
    
    if build_status["running"]:
        return "⚠️ 索引构建正在进行中，请等待完成..."
    
    # 重置状态
    build_status = {"running": False, "progress": "", "error": None, "result": None}
    
    # 启动后台线程
    thread = threading.Thread(
        target=build_index_worker,
        args=(clear_first,),
        daemon=True
    )
    thread.start()
    
    return "🚀 索引构建已启动，请查看下方进度..."


def get_progress():
    """获取当前进度"""
    if build_status["running"]:
        return build_status["progress"] or "处理中..."
    elif build_status["error"]:
        return f"❌ 错误: {build_status['error']}"
    elif build_status["result"]:
        return build_status["result"]
    else:
        return "等待开始..."


# Gradio 界面
with gr.Blocks(title="Weaviate 索引构建工具") as app:
    gr.Markdown("""
    # 🔍 Weaviate 索引构建工具
    
    在 Hugging Face Space 上使用 **免费** Hugging Face embedding，并直接上传到 Weaviate Cloud。
    
    ## 配置要求
    
    请在 **Settings → Secrets** 中添加以下环境变量：
    - `WEAVIATE_URL`: Weaviate Cloud REST 地址
    - `WEAVIATE_API_KEY`: Weaviate API Key
    - `WEAVIATE_COLLECTION`: Collection 名称（默认: GenAICourses）
    - `EMBEDDING_PROVIDER`: huggingface（免费，默认）或 openai（需 OPENAI_API_KEY）
    
    ## 使用步骤
    
    1. 确保已将 `GENAI COURSES` 文件夹上传到 Space 根目录
    2. 点击下方按钮开始构建索引
    3. 等待构建完成（可能需要几分钟）
    """)
    
    with gr.Row():
        clear_first = gr.Checkbox(
            label="清空旧索引后重建",
            value=True,
            info="如果勾选，会先删除旧的 collection 再重建"
        )
        build_btn = gr.Button("🚀 开始构建索引", variant="primary", size="lg")
    
    progress_output = gr.Textbox(
        label="构建进度",
        lines=10,
        interactive=False,
        value="等待开始..."
    )
    
    # 初始加载时显示进度
    app.load(
        fn=get_progress,
        inputs=[],
        outputs=progress_output,
    )
    
    build_btn.click(
        fn=start_build,
        inputs=[clear_first],
        outputs=progress_output,
    )


if __name__ == "__main__":
    app.launch()