123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350 |
- package main
- import (
- "context"
- "fmt"
- "github.com/importcjj/sensitive"
- "go.mongodb.org/mongo-driver/bson/primitive"
- "google.golang.org/grpc"
- "gopkg.in/olivere/elastic.v1"
- "gopkg.in/yaml.v2"
- "io/ioutil"
- "log"
- "math/big"
- "net"
- "net/http"
- "regexp"
- "sensitiveWords.udp/proto_grpc"
- "sensitiveWords.udp/util"
- "strconv"
- "strings"
- )
- const (
- YAMLFILE = "./server.yaml"
- )
- var reg_alias = regexp.MustCompile("(税务局|工商行政管理局|文化广播电视新闻出版局|外国专家局|" +
- "中医药管理局|市场监督管理局|广播电视局|医疗保障局|机关事务管理局|粮食和物资储备局|" +
- "监狱管理局|畜牧兽医局|食品药品监督管理局|城市管理行政执法局|城市管理局|国家保密局|密码管理局|" +
- "地方金融监督管理局|住房保障和房屋管理局|质量技术监督局|人力资源与社会保障局|公路管理局|国土资源局|" +
- "卫生和计划生育局|民事政务局|公众安全局|交通管理局|人力资源和社会保障局|劳动和社会保障局|" +
- "住房和城乡建设局|就业服务局|文物管理局|环境保护局|粮食和物资储备局|教育体育局|" +
- "体育局|教育局|招商局|农业局|农机局|水务局|林业局|财政局|审计局|统计局|商务局)$")
- var reglen *regexp.Regexp = regexp.MustCompile("^(.{1,5}|.{40,})$")
- var strReg *regexp.Regexp = regexp.MustCompile("^(.{0,3}工程队|.{0,3}总公司|_+|.{0,2}设备安装公司|.{0,2}装[饰修潢]公司|.{0,2}开发公司|.{0,4}有限公司|.{0,4}有限责任公司|.{0,4}设计院|建筑设计研?究?院|省文物考古研究所|经济开发区|省.*|镇人民政府|.{0,2}服务公司|" +
- ".{0,2}工程质量监督站|.{0,3}经[营销]部|.{0,3}事务所|.{0,4}工程公司|.{0,4}责任公司|.*勘测|.{0,4}研究院|.*能源建|.{0,2}安装工程|.*[市省]{1}|.{0,4}中心|.*区.?|" +
- ".{0,3}税务局|.{0,3}财政局|.{0,3}商行|.{0,2}公安处|.{0,2}测绘院|.{0,3}开发|.{0,2}建设局|.{0,2}经销部|.{0,3}委员会|.{0,2}分公司|.{0,2}管理站|.{0,2}事务管理局|" +
- ".*资料|.{0,2}办公用品.{1,2}|.*唯亭|.*设备|.+安装|.{0,2}技术服务|市.+[台院社局司]|城?区.+[府局室院]|县.+[院台局]|.{0,2}发展公司|经济技术开发|" +
- "发展和改革局|贵州有色地质|铝塑门窗加工|生产力促进中心|特殊普通合伙|工业集团公司|人民调解协会|人民政府办公厅|机电设备公司|房地产开发有限公司|.{0,4}商店|中等专业学校|" +
- "农村信用联社|.{0,4}经营部|.{0,4}销售部|驾驶员培训学校|.{2}县.{2}镇|保安服务总公司|住房和城乡建设局|地产评估事务所|生产资料门市部|×+|.{0,3}[0-9]{15}|.*[0-9]+|.*路|.*无字号名称.*|.*车|.*[,,]{1}.*|.*个体工商户|.*运输户)$")
- //非中文开头...
- var unstart_strReg *regexp.Regexp = regexp.MustCompile("^([\u4e00-\u9fa5])")
- //开头
- var start_strReg *regexp.Regexp = regexp.MustCompile("^([a-zA-Z]{1,2}[\u4e00-\u9fa5]{6,}|省|市|县|区|业绩|资格|中标|项目|预算单位)")
- //结尾
- var end_strReg *regexp.Regexp = regexp.MustCompile("(\\.|\\.\\.|餐馆|店|腻子|肉庄|画社|美发屋|发廊|网吧|网咖|零售点|新街|包子铺|奶茶铺|(株)|先生|女士|小姐|" +
- "资格|业绩|中标|项目|预算单位|摊位号|号|厅|室|部|点|馆|场|厂|床|所|处|站|行|中心|合作社|ATMS|" +
- "吧|楼|摊|摊位|廊|茶社|坊|圃|汤锅|园|民宿|美容院|房|排挡|府|庄|栈|队|批发|苑|养殖户|棋牌|农家乐|货运|" +
- "城|社|基地|会|服务|娱乐|种植|百货|汽修|农家菜|亭|小吃|快餐|粮库|卫生院|书画院|面|门窗|鸡排|屋|橱|堂|肉铺|服务|服饰|/*)$")
- //包含
- var con_strReg *regexp.Regexp = regexp.MustCompile("(\\?|?|%|代码标识|删除|错误|吊销|注销|发起人|待清理|&#|护照号|身份证号|" +
- "法人| |国家拨入|借款|积累资金|单位自有|认股人|--|、|&|`|美元|[\u4e00-\u9fa5]{2,6}·[\u4e00-\u9fa5]{2,6})|" +
- "[a-zA-Z]{5,}")
- var uncon_strReg *regexp.Regexp = regexp.MustCompile("(园|政府|集团|公司|有限|合伙|企|院|学|局|处)")
- var YamlConfig YAMLConfig
- var MixDataMgo *util.MongodbSim
- var Filter *sensitive.Filter
- var es_type, es_index string
- var Client_Es *elastic.Client
- var data_mgo *MongodbSim
- func init() {
- yamlFile, err := ioutil.ReadFile(YAMLFILE)
- if err != nil {
- log.Fatalln("load conf error")
- }
- err = yaml.Unmarshal(yamlFile, &YamlConfig)
- if err != nil {
- fmt.Println(err.Error())
- }
- log.Printf("%#v", YamlConfig)
- MixDataMgo = &util.MongodbSim{
- MongodbAddr: YamlConfig.MixdataMgoAddr,
- Size: YamlConfig.MongodbPoolSize,
- DbName: YamlConfig.DbName,
- UserName: YamlConfig.UserName,
- PassWord: YamlConfig.PassWord,
- }
- MixDataMgo.InitPool()
- data_mgo = &MongodbSim{
- MongodbAddr: "192.168.3.207:27092",
- DbName: "zhengkun",
- Size: 10,
- UserName: "",
- Password: "",
- }
- data_mgo.InitPool()
- Client_Es ,_= elastic.NewClient(http.DefaultClient, "http://192.168.3.11:9800")
- es_type, es_index = "azktest","azktest"
- }
- func main() {
- //测试
- temporaryTest()
- return
- if YamlConfig.IsAddTask==0{
- initSensitiveWordsData() //初始化敏感词数据
- }else {
- go addTaskSensitiveWordsData() //增量-改配置文件
- }
- lis, err := net.Listen("tcp", YamlConfig.Port)
- if err != nil {
- log.Fatalf("failed to listen: %v", err)
- }
- s := grpc.NewServer()
- proto_grpc.RegisterSensitiveWordsServer(s, &server{})
- log.Println("server start:", YamlConfig.Port)
- if err := s.Serve(lis); err != nil {
- log.Fatalf("failed to serve: %v", err)
- }
- }
- //协议方法---等
- type server struct {
- proto_grpc.SensitiveWordsServer
- }
- func (s *server) Search(ctx context.Context, in *proto_grpc.Request) (*proto_grpc.ResultSensitiveWords, error) {
- text := in.GetText()
- log.Println(text)
- findAll := Filter.FindAll(text)
- return &proto_grpc.ResultSensitiveWords{SensitiveWords: strings.Join(findAll, ",")}, nil
- }
- type YAMLConfig struct {
- MixdataMgoAddr string `yaml:"mixdataMgoAddr"`
- UserName string `yaml:"userName"`
- PassWord string `yaml:"passWord"`
- DbName string `yaml:"dbName"`
- MongodbPoolSize int `yaml:"mongodbPoolSize"`
- TaskGteId string `yaml:"taskGteId"`
- TaskLteId string `yaml:"taskLteId"`
- IsAddTask int `yaml:"isAddTask"`
- Port string `yaml:"port"`
- }
- //其他方法
- func StringTOBsonId(id string) primitive.ObjectID {
- objectId, _ := primitive.ObjectIDFromHex(id)
- return objectId
- }
- func BsonTOStringId(id interface{}) string {
- return id.(primitive.ObjectID).Hex()
- }
- func toMegaBytes(bytes uint64) float64 {
- return float64(bytes) / 1024 / 1024
- }
- func IntAll(num interface{}) int {
- return IntAllDef(num, 0)
- }
- func Int64All(num interface{}) int64 {
- if i, ok := num.(int64); ok {
- return int64(i)
- } else if i0, ok0 := num.(int32); ok0 {
- return int64(i0)
- } else if i1, ok1 := num.(float64); ok1 {
- return int64(i1)
- } else if i2, ok2 := num.(int); ok2 {
- return int64(i2)
- } else if i3, ok3 := num.(float32); ok3 {
- return int64(i3)
- } else if i4, ok4 := num.(string); ok4 {
- i64, _ := strconv.ParseInt(i4, 10, 64)
- //in, _ := strconv.Atoi(i4)
- return i64
- } else if i5, ok5 := num.(int16); ok5 {
- return int64(i5)
- } else if i6, ok6 := num.(int8); ok6 {
- return int64(i6)
- } else if i7, ok7 := num.(*big.Int); ok7 {
- in, _ := strconv.ParseInt(fmt.Sprint(i7), 10, 64)
- return int64(in)
- } else if i8, ok8 := num.(*big.Float); ok8 {
- in, _ := strconv.ParseInt(fmt.Sprint(i8), 10, 64)
- return int64(in)
- } else {
- return 0
- }
- }
- func Float64All(num interface{}) float64 {
- if i, ok := num.(float64); ok {
- return float64(i)
- } else if i0, ok0 := num.(int32); ok0 {
- return float64(i0)
- } else if i1, ok1 := num.(int64); ok1 {
- return float64(i1)
- } else if i2, ok2 := num.(int); ok2 {
- return float64(i2)
- } else if i3, ok3 := num.(float32); ok3 {
- return float64(i3)
- } else if i4, ok4 := num.(string); ok4 {
- in, _ := strconv.ParseFloat(i4, 64)
- return in
- } else if i5, ok5 := num.(int16); ok5 {
- return float64(i5)
- } else if i6, ok6 := num.(int8); ok6 {
- return float64(i6)
- } else if i6, ok6 := num.(uint); ok6 {
- return float64(i6)
- } else if i6, ok6 := num.(uint8); ok6 {
- return float64(i6)
- } else if i6, ok6 := num.(uint16); ok6 {
- return float64(i6)
- } else if i6, ok6 := num.(uint32); ok6 {
- return float64(i6)
- } else if i6, ok6 := num.(uint64); ok6 {
- return float64(i6)
- } else if i7, ok7 := num.(*big.Float); ok7 {
- in, _ := strconv.ParseFloat(fmt.Sprint(i7), 64)
- return float64(in)
- } else if i8, ok8 := num.(*big.Int); ok8 {
- in, _ := strconv.ParseFloat(fmt.Sprint(i8), 64)
- return float64(in)
- } else {
- return 0
- }
- }
- func IntAllDef(num interface{}, defaultNum int) int {
- if i, ok := num.(int); ok {
- return int(i)
- } else if i0, ok0 := num.(int32); ok0 {
- return int(i0)
- } else if i1, ok1 := num.(float64); ok1 {
- return int(i1)
- } else if i2, ok2 := num.(int64); ok2 {
- return int(i2)
- } else if i3, ok3 := num.(float32); ok3 {
- return int(i3)
- } else if i4, ok4 := num.(string); ok4 {
- in, _ := strconv.Atoi(i4)
- return int(in)
- } else if i5, ok5 := num.(int16); ok5 {
- return int(i5)
- } else if i6, ok6 := num.(int8); ok6 {
- return int(i6)
- } else if i7, ok7 := num.(*big.Int); ok7 {
- in, _ := strconv.Atoi(fmt.Sprint(i7))
- return int(in)
- } else if i8, ok8 := num.(*big.Float); ok8 {
- in, _ := strconv.Atoi(fmt.Sprint(i8))
- return int(in)
- } else {
- return defaultNum
- }
- }
- func ObjToString(old interface{}) string {
- if nil == old {
- return ""
- } else {
- r, _ := old.(string)
- return r
- }
- }
- func ObjToStringDef(old interface{}, defaultstr string) string {
- if nil == old {
- return defaultstr
- } else {
- r, _ := old.(string)
- if r == "" {
- return defaultstr
- }
- return r
- }
- }
- //对象数组转成string数组
- func ObjArrToStringArr(old []interface{}) []string {
- if old != nil {
- new := make([]string, len(old))
- for i, v := range old {
- new[i] = v.(string)
- }
- return new
- } else {
- return nil
- }
- }
- //对象数组转成map数组
- func ObjArrToMapArr(old []interface{}) []map[string]interface{} {
- if old != nil {
- new := make([]map[string]interface{}, len(old))
- for i, v := range old {
- new[i] = v.(map[string]interface{})
- }
- return new
- } else {
- return nil
- }
- }
- //map数组转成对象数组
- func MapArrToObjArr(old []map[string]interface{}) []interface{} {
- if old != nil {
- new := make([]interface{}, len(old))
- for i, v := range old {
- new[i] = v
- }
- return new
- } else {
- return nil
- }
- }
|