dsProject/dsRag/Tools/T2_ImportTxt.py

import datetime
import logging

from elasticsearch import Elasticsearch
from tqdm import tqdm

from Config.Config import ES_CONFIG
from Util.EmbeddingUtil import text_to_embedding

# 在文件开头添加logger配置
logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)

# 创建控制台handler并设置格式
handler = logging.StreamHandler()
formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
handler.setFormatter(formatter)
logger.addHandler(handler)

def split_paragraphs(text):
    """按段落分割文本"""
    # 按两个换行符分割段落
    paragraphs = [p.strip() for p in text.split('\n\n') if p.strip()]
    return paragraphs

def save_to_es(text):
    """保存向量化文本和原始文本到ES"""
    vector = text_to_embedding(text)
    
    if vector is None:
        logger.warning(f"跳过无法生成向量的文本: {text}")
        return
    
    doc = {
        'text': text,
        'vector': vector,
        'timestamp': datetime.datetime.now().isoformat(),
        'analyzer': 'ik_smart'
    }
    
    try:
        es.index(index='knowledge_base', body=doc)
        es.index(index='raw_texts', body={'raw_text': text})
    except Exception as e:
        logger.error(f"保存文本到ES失败: {e}")

def process_file(file_path):
    """处理文本文件"""
    with open(file_path, 'r', encoding='utf-8') as f:
        content = f.read()
    
    paragraphs = split_paragraphs(content)
    
    for paragraph in tqdm(paragraphs, desc='处理进度', unit='段'):
        save_to_es(paragraph)
    
    print(f"\n处理完成，共保存{len(paragraphs)}个段落")

if __name__ == '__main__':
    es = Elasticsearch(
        hosts=[ES_CONFIG['hosts']],
        basic_auth=ES_CONFIG['basic_auth'],
        verify_certs=ES_CONFIG['verify_certs'],
        ssl_show_warn=ES_CONFIG['ssl_show_warn']
    )
    
    file_path = '../Txt/人口变化趋势对云南教育的影响.txt'
    process_file(file_path)