# -*- coding: utf-8 -*-
"""
文本处理器集合
每个处理器签名：fn(raw: Any, data: IOData) -> Any
可自由组合成处理链，由 IODispatcher 顺序调用
"""
import logging
import re
import time

from uas.io.base_io import IOData

logger = logging.getLogger(__name__)


def processor_clean_text(raw: str, data: IOData) -> str:
    """清洗文本：去除多余空白、特殊字符"""
    result = re.sub(r'\s+', ' ', str(raw)).strip()
    result = re.sub(r'[^\w\s\u4e00-\u9fff.,!?，。！？、；：""''（）【】]', '', result)
    logger.debug(f"[clean] {raw!r} -> {result!r}")
    return result


def processor_to_upper(raw: str, data: IOData) -> str:
    """英文转大写"""
    return raw.upper()


def processor_add_timestamp(raw: str, data: IOData) -> str:
    """追加时间戳"""
    ts = time.strftime("%Y-%m-%d %H:%M:%S")
    return f"[{ts}] {raw}"


def processor_word_count(raw: str, data: IOData) -> str:
    """统计字数并追加到 metadata"""
    words = len(raw.split())
    chars = len(raw)
    data.metadata["word_count"] = words
    data.metadata["char_count"] = chars
    logger.debug(f"字数统计: words={words} chars={chars}")
    return raw   # 不修改内容，仅更新 metadata


def processor_keyword_extract(raw: str, data: IOData) -> str:
    """简单关键词提取（示例：提取长度>3的词）"""
    words    = re.findall(r'\b\w{4,}\b', raw)
    keywords = list(set(words))[:10]
    data.metadata["keywords"] = keywords
    logger.debug(f"关键词: {keywords}")
    return raw


def processor_sensitive_filter(raw: str, data: IOData) -> str:
    """敏感词过滤（示例）"""
    sensitive = ["badword1", "badword2"]
    result = raw
    for word in sensitive:
        result = result.replace(word, "*" * len(word))
    return result


def processor_simulate_nlp(raw: str, data: IOData) -> dict:
    """
    模拟 NLP 处理（实际可替换为 transformers / spacy 等）
    返回结构化结果
    """
    import time
    time.sleep(0.05)   # 模拟耗时
    return {
        "original": raw,
        "length":   len(raw),
        "summary":  raw[:50] + ("..." if len(raw) > 50 else ""),
        "intent":   "query" if "?" in raw or "？" in raw else "statement",
        "lang":     "zh" if re.search(r'[\u4e00-\u9fff]', raw) else "en",
    }


# ─────────────────────────────────────────────────────────────────
# 处理器工厂：预设处理链
# ─────────────────────────────────────────────────────────────────

PIPELINE_PRESETS = {
    "basic": [
        processor_clean_text,
        processor_sensitive_filter,
        processor_word_count,
    ],
    "full": [
        processor_clean_text,
        processor_sensitive_filter,
        processor_word_count,
        processor_keyword_extract,
        processor_simulate_nlp,
    ],
    "speech": [
        processor_clean_text,
        processor_add_timestamp,
    ],
}