# from langchain_openai import OpenAIEmbeddings
# from utils import CheckResult, StandardType, load_standard_name
#
# standard_program_name_list = load_standard_name('./standard_data/standard_program.txt')
#
# params =  {'model': 'bge-large-zh-v1.5',
#            'openai_api_base': 'http://218.23.122.14:63015/v1-openai/',
#            'openai_api_key': 'gpustack_baacebfd27bb3d01_092ce528ae05cb7d05acb052e6490090',
#            'openai_proxy': ''}
#
# try:
#     embedding = OpenAIEmbeddings(**params)
#     result = embedding.embed_documents(standard_program_name_list,chunk_size=500)
#
#     print(f"mbedding.embed_documents 结果：{result}")
#
#
# except Exception as e:
#     print(f"failed to create Embeddings for model. {e}")


from langchain_openai import OpenAIEmbeddings
from utils import CheckResult, StandardType, load_standard_name
from sklearn.metrics.pairwise import cosine_similarity
import numpy as np

# 加载标准项目部名称列表
standard_program_name_list = load_standard_name('./standard_data/standard_program.txt')
# 模型参数
params =  {'model': 'bge-large-zh-v1.5',
           'openai_api_base': 'http://127.0.0.1:9997/v1',
           'openai_api_key': 'EMPTY',
           'openai_proxy': ''}
# 创建嵌入模型
embedding = OpenAIEmbeddings(**params)

# 获取标准项目部名称的嵌入向量
standard_embeddings = embedding.embed_documents(standard_program_name_list, chunk_size=500)


def fuzzy_match(query):
    try:
        from sklearn.metrics.pairwise import cosine_similarity
        import numpy as np
        # 查询名称
        query_embedding = embedding.embed_query(query)

        # 计算相似度
        similarities = cosine_similarity([query_embedding], standard_embeddings)[0]

        # 找到最相似的项目部名称
        most_similar_index = np.argmax(similarities)
        most_similar_name = standard_program_name_list[most_similar_index]
        print(f"输入名称: {query}")
        print(f"最相似的项目部名称: {most_similar_name}")
        print(f"相似度: {similarities[most_similar_index]:.4f}")
        return most_similar_name, similarities[most_similar_index]
    except Exception as e:
        print(f"相似性判断错误{e}")

# try:
#     # 查询名称
#     query = "定西第一项目部"
#     query_embedding = embedding.embed_query(query)
#
#     # 计算相似度
#     similarities = cosine_similarity([query_embedding], standard_embeddings)[0]
#
#     # 找到最相似的项目部名称
#     most_similar_index = np.argmax(similarities)
#     most_similar_name = standard_program_name_list[most_similar_index]
#
#     print(f"输入名称: {query}")
#     print(f"最相似的项目部名称: {most_similar_name}")
#     print(f"相似度: {similarities[most_similar_index]:.4f}")
#
# except Exception as e:
#     print(f"Failed to create embeddings or compute similarity: {e}")

match_program, match_possibility = fuzzy_match("第一项目部定西")
print(f"fuzzy_match program result:{match_program}, {match_possibility}")