simple-memo/utils/sensitive.go

// utils/sensitive.go
// 敏感词过滤工具
package utils

import (
	"embed"
	"strings"
	"sync"

	"github.com/importcjj/sensitive"
)

//go:embed sensitive_words.txt
var sensitiveWordsFS embed.FS

var (
	filter     *sensitive.Filter
	filterOnce sync.Once
)

// GetSensitiveFilter 获取敏感词过滤器单例
func GetSensitiveFilter() *sensitive.Filter {
	filterOnce.Do(func() {
		filter = sensitive.New()
		loadDefaultWords()
	})
	return filter
}

// loadDefaultWords 加载默认敏感词库
func loadDefaultWords() {
	// 从嵌入的文件系统读取
	data, err := sensitiveWordsFS.ReadFile("sensitive_words.txt")
	if err != nil {
		// 如果文件不存在，使用内置的基础词库
		loadBuiltinWords()
		return
	}

	words := strings.Split(string(data), "\n")
	for _, word := range words {
		word = strings.TrimSpace(word)
		if word != "" && !strings.HasPrefix(word, "#") {
			filter.AddWord(word)
		}
	}
}

// loadBuiltinWords 加载内置基础敏感词
func loadBuiltinWords() {
	// 基硎敏感词列表
	builtinWords := []string{
		// 政治相关
		"习近平", "江泽民", "胡锦涛", "温家宝", "李克强",
		"法轮功", "六四", "天安门", "台独", "藏独", "疆独",
		// 色情相关
		"色情", "黄色", "裸体", "做爱", "性交", "强奸", "乱伦",
		// 暴力相关
		"杀人", "自杀", "爆炸", "恐怖", "炸弹", "袭击",
		// 赌博相关
		"赌博", "赌钱", "博彩", "六合彩",
		// 毒品相关
		"毒品", "吸毒", "贩毒", "冰毒", "海洛因",
		// 侮辱性词汇
		"傻逼", "操你", "妈的", "王八蛋", "畜生",
	}

	for _, word := range builtinWords {
		filter.AddWord(word)
	}
}

// FilterSensitive 过滤敏感词，替换为 *
func FilterSensitive(text string) string {
	f := GetSensitiveFilter()
	return f.Replace(text, '*')
}

// ContainsSensitive 检查是否包含敏感词
func ContainsSensitive(text string) bool {
	f := GetSensitiveFilter()
	has, _ := f.FindIn(text)
	return has
}

// FindAllSensitive 查找所有敏感词
func FindAllSensitive(text string) []string {
	f := GetSensitiveFilter()
	return f.FindAll(text)
}

// ValidateContent 验证内容，返回是否通过和第一个敏感词
func ValidateContent(text string) (bool, string) {
	f := GetSensitiveFilter()
	return f.Validate(text)
}

// AddSensitiveWord 动态添加敏感词
func AddSensitiveWord(word string) {
	f := GetSensitiveFilter()
	f.AddWord(word)
}

// AddSensitiveWords 批量添加敏感词
func AddSensitiveWords(words []string) {
	f := GetSensitiveFilter()
	for _, word := range words {
		f.AddWord(word)
	}
}