from langchain.vectorstores import Qdrant
from langchain.embeddings import HuggingFaceEmbeddings
from qdrant_client import QdrantClient
import gradio as gr

def process_text(input_text, top_k):
    embeddings = HuggingFaceEmbeddings(model_name="intfloat/multilingual-e5-base")
    client = QdrantClient(
        path="./local_qdrant", prefer_grpc=True
    )
    db = Qdrant(client=client, embeddings=embeddings, collection_name="qa_data")

    query = input_text

    all_answers = []
    docs = db.similarity_search_with_score(query=query, k=top_k)
    for i in docs:
        doc, score = i
        print({"score": score, "content": doc.page_content, "metadata": doc.metadata} )
        all_answers.append(doc.metadata["source"])

    return "\n***\\n".join(all_answers)

CSS ="""
.contain { display: flex; flex-direction: column; }
.gradio-container { height: 100vh !important; }
#component-0 { height: 100%; }
#textbox { flex-grow: 1; overflow: auto; resize: vertical; }
.secondary {background-color: #6366f1; }
"""
#with gr.Blocks() as demo:
with gr.Blocks(theme=gr.themes.Monochrome(radius_size=gr.themes.sizes.radius_sm)) as demo:
    with gr.Row():
        gr.Markdown("# 裁定検索")
    with gr.Row():
        output = gr.TextArea(
            elem_id="検索結果",
            label="検索結果",
        )
    with gr.Row():
        input = gr.Textbox(
            label="質問",
            placeholder="芸魔龍王アメイジンの出た時の効果は、後から出たクリーチャーも影響しますか",
            lines=3,
        )
    with gr.Row():
        submit = gr.Button(value="検索", variant="secondary").style(full_width=True)
        top_k = gr.Slider(1, 10, label="表示数", step=1, value=5, interactive=True)
    
    submit_click_event = submit.click(fn=process_text, inputs=[input, top_k], outputs=output)

demo.launch()
# demo.queue(max_size=128, concurrency_count=48).launch(debug=True, server_name="0.0.0.0", server_port=7860)