test_clean.py

import re
def clean_chinese(text: str) -> str:
    cleaned = re.sub(r'[a-zA-Z\'\-]{3,}', '', text)
    cleaned = re.sub(r'[^\u4e00-\u9fff\u3000-\u303f\uff00-\uffef，。！？、；：""''（）—…\s]', '', cleaned)
    cleaned = re.sub(r'\s{2,}', ' ', cleaned)
    cleaned = cleaned.strip()
    
    cleaned = re.sub(r'，\s*(并且|而且|但是|可是|然而|而|因此|所以|于是|因为|由于|虽然)', r'。\1', cleaned)
    
    segments = cleaned.split('，')
    result = []
    current_len = 0
    for i, seg in enumerate(segments):
        result.append(seg)
        current_len += len(seg)
        if i < len(segments) - 1:
            if current_len > 12:
                result.append('。')
                current_len = 0
            else:
                result.append('，')
    cleaned = "".join(result)
    
    cleaned = re.sub(r'。+', '。', cleaned)
    cleaned = cleaned.replace('。，', '。').replace('，。', '。')
    if cleaned and cleaned[-1] not in ['。', '！', '？', '”']:
        cleaned += '。'

    if len(cleaned) < 5:
        return ""
    return cleaned

test_str = "这是一个测试句子，看看长句能不能被正确打断，并且这里的逻辑会不会导致空字符串。"
print("Result:", clean_chinese(test_str))