12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079 |
- // front
- package front
- import (
- "fmt"
- "github.com/go-xweb/httpsession"
- "github.com/go-xweb/xweb"
- "github.com/lauyoume/gopinyin"
- "github.com/tealeg/xlsx"
- "io/ioutil"
- "log"
- "mongodb"
- "os"
- qu "qfw/util"
- "regexp"
- "sort"
- "spider"
- util "spiderutil"
- "strconv"
- "strings"
- "sync"
- task "taskManager"
- "time"
- u "util"
- )
- type Front struct {
- *xweb.Action
- login xweb.Mapper `xweb:"/"`
- logout xweb.Mapper `xweb:"/center/logout"` //退出
- loadIndex xweb.Mapper `xweb:"/center"` //控制中心
- spidernew xweb.Mapper `xweb:"/center/spider"` //爬虫新建
- reg xweb.Mapper `xweb:"/center/reg"` //爬虫注册
- assign xweb.Mapper `xweb:"/center/user/assign"` //分配爬虫
- auditExport xweb.Mapper `xweb:"/center/user/auditexport"` //导出审核日志
- loadSpider xweb.Mapper `xweb:"/center/spider/edit/(.*)"` //爬虫加载
- viewSpider xweb.Mapper `xweb:"/center/spider/view/(.*)"` //爬虫查看
- downSpider xweb.Mapper `xweb:"/center/spider/download/(.*)"` //爬虫下载
- upState xweb.Mapper `xweb:"/center/spider/upstate"` //爬虫状态更新
- assort xweb.Mapper `xweb:"/center/spider/assort"` //审核人员分类(无发布、需登录、无法处理、需删除)
- batchShelves xweb.Mapper `xweb:"/center/spider/batchShelves"` //批量上下架
- checktime xweb.Mapper `xweb:"/center/spider/checktime"` //爬虫核对
- disables xweb.Mapper `xweb:"/center/spider/disable"` //批量作废
- changeEvent xweb.Mapper `xweb:"/center/changeEvent"` //节点更新
- getJson xweb.Mapper `xweb:"/center/spider/json"` //
- //delRedis xweb.Mapper `xweb:"/center/spider/delRedis"` //清理Redis
- updateESP xweb.Mapper `xweb:"/center/spider/updateesp"` //修改爬虫的节点/状态/平台
- updatePendState xweb.Mapper `xweb:"/center/spider/updatePendState"` //更新爬虫挂起状态
- tagCode xweb.Mapper `xweb:"/center/spider/tagcode"` //标记爬虫
- spiderRemark xweb.Mapper `xweb:"/center/spider/spiderremark"` //标记爬虫
- spiderModel xweb.Mapper `xweb:"/center/model"` //获取补充模型
- runStep xweb.Mapper `xweb:"/center/run"` //方法测试
- spiderPass xweb.Mapper `xweb:"/center/spider/pass"` //整体测试
- runPinYin xweb.Mapper `xweb:"/center/runpy"` //获取拼音
- saveStep xweb.Mapper `xweb:"/center/save"` //爬虫保存
- saveJs xweb.Mapper `xweb:"/center/save/js"` //保存js
- loadModel xweb.Mapper `xweb:"/center/gmodel/(.*)"` //加载模型
- importdata xweb.Mapper `xweb:"/center/importdata"` //导入爬虫列表页面
- importLua xweb.Mapper `xweb:"/center/importlua"` //导入爬虫
- oldedit xweb.Mapper `xweb:"/center/oldedit"` //老文件编辑
- findName xweb.Mapper `xweb:"/center/findname"` //即时查询名称
- checkrepeat xweb.Mapper `xweb:"/center/spider/isrepeat"` //脚本代码判重
- heart xweb.Mapper `xweb:"/center/heart"` //心跳监控
- spiderCopy xweb.Mapper `xweb:"/center/spider/copy"` //补采复制爬虫
- spiderSplitCopy xweb.Mapper `xweb:"/center/spider/splitcopy"` //拆分复制爬虫
- spiderCloseErr xweb.Mapper `xweb:"/center/spider/closeerr"` //删除错误爬虫
- Base Base
- OtherBase OtherBase
- Step1 Step1
- Step2 Step2
- Step3 Step3
- StepRe3 StepRe3
- U U
- luaList xweb.Mapper `xweb:"/center/lualist.html"` //脚本管理
- user xweb.Mapper `xweb:"/center/user.html"` //用户管理
- delUser xweb.Mapper `xweb:"/center/user/del"` //删除用户
- updateUser xweb.Mapper `xweb:"/center/user/updateUser"` //修改用户信息
- checkUsenamer xweb.Mapper `xweb:"/center/user/checkUsenamer"` //校验用户名的唯一性
- checkEmail xweb.Mapper `xweb:"/center/user/checkEmail"` //校验邮箱的唯一性
- saveNewUser xweb.Mapper `xweb:"/center/user/saveNewUser"` //添加用户
- getCity xweb.Mapper `xweb:"/center/getCity"` //获取城市
- //补采信息
- supplementDayList xweb.Mapper `xweb:"/center/supplement/daylist"` //日补采列表
- supplementWeekList xweb.Mapper `xweb:"/center/supplement/weeklist"` //周补采列表
- }
- const Sp_state_0, Sp_state_1, Sp_state_2, Sp_state_3, Sp_state_4, Sp_state_5, Sp_state_6, Sp_state_7, Sp_state_8, Sp_state_9, Sp_state_10 = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 //0待完成,1待审核,2打回,3发布,4作废,5已上架,6已下架,7无发布,8需登录,9转python,10已删除
- var LuaReg = regexp.MustCompile(`(:(find|match)|string\.|com\.getPureContent)`)
- var spinfos sync.Map = sync.Map{}
- var SessMap map[string]*httpsession.Session
- var AutoTpl map[string]interface{}
- var Mails *util.Mail
- var Reg = regexp.MustCompile(`(http|https)://([\w]+\.)+[\w]+`)
- var ListFilterReg = regexp.MustCompile("关键词过滤")
- var ProjectHrefReg = regexp.MustCompile("projecthref")
- var Transfercode map[string]interface{}
- // 控制中心
- func (f *Front) LoadIndex() {
- auth := qu.IntAll(f.GetSession("auth"))
- if f.Method() == "POST" {
- start, _ := f.GetInteger("start")
- limit, _ := f.GetInteger("length")
- draw, _ := f.GetInteger("draw")
- searchStr := f.GetString("search[value]")
- event, _ := f.GetInteger("taskEvent") //节点
- //searchN := strings.Replace(searchStr, " ", "", -1)
- //search := strings.Replace(searchN, "\n", "", -1)
- search := strings.TrimSpace(searchStr)
- state, _ := f.GetInteger("state")
- urgency, _ := f.GetInteger("urgency") //节点
- platform := f.GetString("platform")
- modifyuser := f.GetString("modifyuser")
- infoformat, _ := f.GetInteger("infoformat") //节点
- query := map[string]interface{}{}
- if event > -1 {
- query["event"] = event
- }
- if urgency > -1 {
- query["urgency"] = urgency
- }
- if platform != "-1" {
- query["platform"] = platform
- }
- if modifyuser != "-1" {
- query["modifyuser"] = modifyuser
- }
- if infoformat > -1 {
- query["infoformat"] = infoformat
- }
- if search != "" {
- query["$or"] = []interface{}{
- map[string]interface{}{"code": map[string]interface{}{"$regex": search}},
- map[string]interface{}{"createuser": map[string]interface{}{"$regex": search}},
- map[string]interface{}{"param_common.1": map[string]interface{}{"$regex": search}},
- }
- }
- if auth == u.Role_Examine { //审核员
- if state > -1 {
- query["state"] = state
- } else {
- query["state"] = Sp_state_1
- }
- } else if auth == u.Role_Dev { //开发员
- if state > -1 {
- query["state"] = state
- }
- query["createuserid"] = f.GetSession("userid")
- } else { //管理员
- if state > -1 {
- query["state"] = state
- }
- }
- sort := `{"%s":%d}`
- orderIndex := f.GetString("order[0][column]")
- orderName := f.GetString(fmt.Sprintf("columns[%s][data]", orderIndex))
- orderType := 1
- if f.GetString("order[0][dir]") != "asc" {
- orderType = -1
- }
- if orderName == "param_common" {
- orderName = orderName + ".1"
- }
- sort = fmt.Sprintf(sort, orderName, orderType)
- page := start / 10
- //log.Println("sort", sort, orderName)
- qu.Debug("query:", query, "sort:", sort)
- //luas := *mgdb.Find("luaconfig", query, sort, list_fields, false, start, limit)
- //count := mgdb.Count("luaconfig", query)
- luas, _ := u.MgoEB.Find("luaconfig", query, sort, map[string]interface{}{"str_list": 0, "str_content": 0}, false, start, limit)
- count := u.MgoEB.Count("luaconfig", query)
- for k, v := range *luas {
- v["num"] = k + 1 + page*10
- if v["modifytime"] != nil {
- v["modifytime"] = time.Unix(v["modifytime"].(int64), 0).Format("2006-01-02 15:04:05")
- } else {
- v["modifytime"] = "-"
- }
- if v["modifyuser"] == nil {
- v["modifytime"] = "-"
- }
- v["encode"] = util.Se.Encode2Hex(fmt.Sprint(v["code"]))
- if v["event"] == nil { //节点
- v["event"] = 0
- }
- user, _ := u.MgoEB.FindOne("user", map[string]interface{}{"s_name": qu.ObjToString(v["createuser"])})
- //v["state"] = LuaStateMap[qu.IntAll(v["state"])]
- v["i_scope"] = (*user)["i_scope"]
- }
- f.ServeJson(map[string]interface{}{"draw": draw, "data": luas, "recordsFiltered": count, "recordsTotal": count})
- } else {
- events := []string{}
- for k, _ := range util.Config.Uploadevents {
- events = append(events, k)
- }
- f.T["modifyusers"] = task.GetModifyUsers()
- sort.Strings(events)
- f.T["events"] = events
- f.Render("index.html", &f.T)
- }
- }
- func (f *Front) Checkrepeat() {
- code := f.GetString("code")
- //one := *mgdb.FindOne("luaconfig", bson.M{"code": code})
- one, _ := u.MgoEB.FindOne("luaconfig", map[string]interface{}{"code": code})
- if len(*one) > 0 {
- f.ServeJson("y")
- } else {
- f.ServeJson("n")
- }
- }
- // 新建
- func (f *Front) Spidernew() error {
- auth := qu.IntAll(f.GetSession("auth"))
- if auth != u.Role_Admin {
- return nil
- }
- copy := f.GetString("copy")
- if copy != "" {
- //one := *mgdb.FindOne("luaconfig", bson.M{"code": copy})
- one, _ := u.MgoEB.FindOne("luaconfig", map[string]interface{}{"code": copy})
- delete((*one), "_id")
- delete((*one), "code")
- base := (*one)["param_common"].([]interface{})
- base[0] = ""
- base[1] = ""
- (*one)["param_common"] = base
- f.T["lua"] = one
- }
- f.T["isflow"] = 1 //新建爬虫时,初始化isflow的值
- f.T["actiontext"] = "新建"
- f.T["restate"] = 4 //此处设置restate=4无意义,只为了页面不报错
- f.T["areas"] = u.Area //
- f.T["citys"] = u.City //
- f.T["provinces"] = u.Province //
- return f.Render("spideredit.html", &f.T)
- }
- // 得到模型
- func (f *Front) SpiderModel() {
- f.ServeJson(util.Config.Model)
- }
- func (f *Front) RunPinYin() {
- word := f.GetString("word")
- str := gopinyin.Convert(word, true)
- f.Write(str)
- }
- type U struct {
- User string
- Name string
- Pwd string
- }
- func (f *Front) Reg() {
- }
- func (f *Front) ImportLua() {
- auth := qu.IntAll(f.GetSession("auth"))
- if auth != u.Role_Admin {
- f.ServeJson("没有权限")
- return
- }
- if f.Method() == "POST" {
- mf, _, err := f.GetFile("xlsx")
- errorinfo := map[string]interface{}{}
- if err == nil {
- binary, _ := ioutil.ReadAll(mf)
- xls, _ := xlsx.OpenBinary(binary)
- sheet := xls.Sheets[0]
- rows := sheet.Rows
- for k, v := range rows {
- if k != 0 {
- cells := v.Cells
- if cells[1].Value != "" {
- code := cells[1].Value
- code = u.SymbolReg.ReplaceAllString(code, "")
- query := map[string]interface{}{"code": cells[1].Value}
- rs, _ := u.MgoEB.FindOne("import", query)
- if len(*rs) > 0 {
- errorinfo[cells[1].Value] = "第" + strconv.Itoa(k) + "行爬虫代码重复,请修改"
- continue
- }
- o := make(map[string]interface{})
- o["name"] = cells[0].Value
- o["code"] = code
- o["channel"] = cells[2].Value
- spiderremark := cells[3].Value
- if spiderremark == "" {
- spiderremark = `采集“` + cells[2].Value + `”栏目(含子栏目)`
- }
- o["spiderremark"] = spiderremark
- //重复域名的网站不再新增爬虫
- href := cells[4].Value
- one, _ := u.MgoEB.FindOne("luaconfig", map[string]interface{}{"href": href})
- if len(*one) > 0 {
- errorinfo[cells[1].Value] = "第" + strconv.Itoa(k) + "行爬虫" + cells[1].Value + ",该栏目已存在,请确认"
- //continue
- }
- o["channeladdr"] = href
- o["author"] = cells[5].Value
- o["timestamp"] = time.Now().Unix()
- o["status"] = 1
- o["next"] = cells[5].Value
- o["event"] = cells[6].Value
- o["incrementevent"] = cells[7].Value
- if cells[8].Value == "是" {
- o["isflow"] = 1
- } else {
- o["isflow"] = 0
- }
- if cells[9].Value == "紧急" {
- o["urgency"] = 1
- } else {
- o["urgency"] = 0
- }
- o["platform"] = cells[10].Value
- o["area"] = cells[11].Value
- o["city"] = cells[12].Value
- o["district"] = cells[13].Value
- weigh, _ := cells[14].Int()
- o["weight"] = weigh
- //爬虫类型
- infoformat, _ := cells[15].Int()
- if infoformat < 1 {
- errorinfo[cells[1].Value] = "第" + strconv.Itoa(k) + "行爬虫" + cells[1].Value + ",缺少爬虫类型信息"
- continue
- }
- o["infoformat"] = infoformat
- //存储表
- coll := cells[16].Value
- if strings.Contains(code, "bidding") {
- errorinfo[cells[1].Value] = "第" + strconv.Itoa(k) + "行爬虫" + cells[1].Value + ",存储表错误"
- continue
- }
- o["coll"] = coll
- //table := cells[6].Value
- //o["table"] = table
- //o["transfercode"] = qu.IntAll(Transfercode[table])
- ok, name := saveLua(o) //保存爬虫
- if ok == false {
- errorinfo[cells[1].Value] = "第" + strconv.Itoa(k) + "行找不到作者,已经过滤"
- } else {
- o["author"] = name
- o["importuser"] = f.GetSession("username")
- u.MgoEB.Save("import", o)
- }
- }
- }
- }
- f.ServeJson(errorinfo)
- } else {
- f.ServeJson(false)
- }
- }
- }
- func saveLua(o map[string]interface{}) (bool, string) {
- AutoTpl["Base.SpiderName"] = o["name"]
- AutoTpl["Base.SpiderCode"] = o["code"]
- AutoTpl["Base.SpiderChannel"] = o["channel"]
- AutoTpl["Base.SpiderTargetChannelUrl"] = o["channeladdr"]
- author := o["author"].(string)
- one, _ := u.MgoEB.FindOne("user", map[string]interface{}{"s_email": author})
- id := mongodb.BsonIdToSId((*one)["_id"])
- if len(*one) == 0 {
- return false, ""
- }
- common := []interface{}{
- AutoTpl["Base.SpiderCode"],
- AutoTpl["Base.SpiderName"],
- AutoTpl["Base.SpiderChannel"],
- AutoTpl["Base.SpiderDownDetailPage"],
- AutoTpl["Base.SpiderStartPage"],
- AutoTpl["Base.SpiderMaxPage"],
- AutoTpl["Base.SpiderRunRate"],
- //AutoTpl["Base.Spider2Collection"],
- //"bidding", //爬虫导入新建默认为bidding
- qu.ObjToString(o["coll"]),
- AutoTpl["Base.SpiderPageEncoding"],
- AutoTpl["Base.SpiderStoreMode"],
- AutoTpl["Base.SpiderStoreToMsgEvent"],
- AutoTpl["Base.SpiderTargetChannelUrl"],
- AutoTpl["Base.SpiderLastDownloadTime"],
- AutoTpl["Base.SpiderIsHistoricalMend"],
- AutoTpl["Base.SpiderIsMustDownload"],
- }
- ptime := []interface{}{
- AutoTpl["Step1.DateFormat"],
- AutoTpl["Step1.Address"],
- AutoTpl["Step1.ContentChooser"],
- }
- list := []interface{}{
- AutoTpl["Step2.Listadd"],
- AutoTpl["Step2.Listadds"],
- AutoTpl["Step2.BlockChooser"],
- AutoTpl["Step2.AddressChooser"],
- AutoTpl["Step2.TitleChooser"],
- AutoTpl["Step2.DateChooser"],
- AutoTpl["Step2.DateFormat"],
- }
- content := []interface{}{
- AutoTpl["Step3.ContentChooser"],
- AutoTpl["Step3.ElementChooser"],
- }
- param := map[string]interface{}{}
- param["param_common"] = common
- //向导模式
- param["param_time"] = ptime
- param["param_list"] = list
- param["param_content"] = content
- param["type_time"] = 0
- param["type_list"] = 0
- param["type_content"] = 0
- //专家模式
- param["str_time"] = ""
- param["str_list"] = ""
- param["str_content"] = ""
- param["comeintime"] = time.Now().Unix()
- param["code"] = o["code"]
- param["site"] = o["name"]
- param["href"] = o["channeladdr"]
- param["channel"] = o["channel"]
- param["createuser"] = (*one)["s_name"]
- param["createuserid"] = id
- param["createuseremail"] = (*one)["s_email"]
- param["modifyuser"] = (*one)["s_name"]
- param["modifyuserid"] = id
- param["modifytime"] = time.Now().Unix()
- param["state"] = 0 //未完成
- if qu.IntAll(o["event"]) > 0 {
- param["event"] = qu.IntAll(o["event"])
- }
- s_model := "bid"
- configModel := util.Config.Model[s_model]
- model := map[string]interface{}{}
- for k, _ := range configModel {
- model[k] = qu.ObjToString(o[k])
- }
- param["model"] = model
- param["next"] = o["next"]
- param["urgency"] = o["urgency"]
- param["isflow"] = o["isflow"]
- param["spidertype"] = "history"
- param["spiderremark"] = o["spiderremark"]
- incrementevent := qu.ObjToString(o["incrementevent"])
- if movevent, ok := util.Config.Uploadevents[incrementevent].(string); ok && movevent != "" {
- param["spidermovevent"] = movevent
- }
- //} else {
- // param["spidermovevent"] = "7700"
- //}
- param["incrementevent"] = qu.IntAll(o["incrementevent"])
- param["platform"] = o["platform"]
- param["weight"] = o["weight"]
- param["infoformat"] = o["infoformat"]
- infoformat := qu.IntAll(o["infoformat"])
- infotype := "招标"
- if infoformat == 2 {
- infotype = "拟建/审批"
- } else if infoformat == 3 {
- infotype = "产权"
- } else if infoformat == 4 {
- infotype = "舆情"
- }
- //默认字段
- param["spidercompete"] = true //2021-11-20后爬虫加此字段(表示新爬虫,剑鱼网站不展示原文)
- param["spiderhistorymaxpage"] = 1 //历史最大页
- param["pendstate"] = 0 //
- param["grade"] = 0 //爬虫难易度(主要用于python爬虫使用)
- //qu.Debug("param---", param)
- ok := spider.SaveSpider(o["code"].(string), param)
- if ok { //保存成功,校验新导入的爬虫对应站点是否存在,否则加站点记录
- if u.MgoEB.Count("site", map[string]interface{}{"site": o["name"]}) == 0 {
- qu.Debug("补充站点信息:", o["name"])
- domain := u.DomainReg.FindString(qu.ObjToString(AutoTpl["Base.SpiderTargetChannelUrl"]))
- if domain != "" {
- domain = u.ReplaceReg.ReplaceAllString(domain, "")
- }
- siteInfo := map[string]interface{}{
- "site": o["name"],
- "domain": domain, //
- "another_name": "",
- "area": qu.ObjToString(model["area"]),
- "city": qu.ObjToString(model["city"]),
- "district": qu.ObjToString(model["district"]),
- "site_type": "",
- "second_type": "",
- "industry": "",
- "p_site": "",
- "s_site": "",
- "remarktime": time.Now().Unix(),
- "event": incrementevent,
- "platform": o["platform"],
- "spider_status": "0/1",
- "updatetime": time.Now().Unix(),
- "delete": false,
- "comeintime": time.Now().Unix(),
- "important": 0,
- "site_status": 1,
- "lasttime": int64(0),
- "site_datanum": 0,
- "period": float32(0),
- "infotype": infotype,
- "sponsor": "",
- "isneedregister": 0,
- "isregistered": 0,
- "special_type": "",
- "account": "",
- "password": "",
- "f_area": "",
- "f_city": "",
- "f_district": "",
- "site_subtype": "",
- "site_toptype": "",
- "type_plate": "",
- }
- u.MgoEB.Save("site", siteInfo)
- }
- }
- return ok, (*one)["s_name"].(string)
- }
- func (f *Front) Importdata() {
- auth := qu.IntAll(f.GetSession("auth"))
- if auth == u.Role_Admin {
- if f.Method() == "GET" {
- f.Render("import.html")
- } else {
- query := map[string]interface{}{}
- searchStr := f.GetString("search[value]")
- search := strings.TrimSpace(searchStr)
- if search != "" {
- query["$or"] = []interface{}{
- map[string]interface{}{"code": map[string]interface{}{"$regex": search}},
- map[string]interface{}{"name": map[string]interface{}{"$regex": search}},
- }
- }
- start, _ := f.GetInteger("start")
- limit, _ := f.GetInteger("length")
- draw, _ := f.GetInteger("draw")
- data, _ := u.MgoEB.Find("import", query, `{"timestamp": -1}`, nil, false, start, limit)
- count := u.MgoEB.Count("import", query)
- f.ServeJson(map[string]interface{}{
- "draw": draw,
- "data": data,
- "recordsFiltered": count,
- "recordsTotal": count,
- })
- }
- } else {
- f.Write("您没有导入脚本的权限")
- }
- }
- func Wlog(name, code, man, manid, types string, content map[string]interface{}) {
- obj := map[string]interface{}{
- "name": name,
- "code": code,
- "man": man,
- "manid": manid,
- "types": types,
- "time": time.Now().Unix(),
- "content": content,
- }
- u.MgoEB.Save("lua_logs", obj)
- }
- func (f *Front) Oldedit() {
- if f.Method() == "GET" {
- f.Render("oldedit.html")
- } else {
- }
- }
- func (f *Front) FindName() {
- words := f.GetString("words")
- if words == "" {
- f.ServeJson(map[string]interface{}{"error": "null"})
- }
- //query := bson.M{"$or": []interface{}{
- // bson.M{"param_common.0": bson.M{"$regex": words}},
- // bson.M{"param_common.1": bson.M{"$regex": words}},
- // bson.M{"createuser": bson.M{"$regex": words}},
- //}, "oldlua": bson.M{"$exists": false}}
- query := map[string]interface{}{
- "$or": []interface{}{
- map[string]interface{}{"param_common.0": map[string]interface{}{"$regex": words}},
- map[string]interface{}{"param_common.1": map[string]interface{}{"$regex": words}},
- map[string]interface{}{"createuser": map[string]interface{}{"$regex": words}},
- },
- "oldlua": map[string]interface{}{
- "$exists": false,
- },
- }
- //rs := *mgdb.Find("luaconfig", query, bson.M{"modifytime": -1}, bson.M{"param_common": 1}, false, -1, -1)
- rs, _ := u.MgoEB.Find("luaconfig", query, map[string]interface{}{"modifytime": -1}, map[string]interface{}{"param_common": 1}, false, -1, -1)
- if len(*rs) > 0 {
- f.ServeJson(map[string]interface{}{"data": (*rs)})
- } else {
- f.ServeJson(map[string]interface{}{"error": "data"})
- }
- }
- // 分配爬虫
- func (f *Front) Assign() {
- auth := qu.IntAll(f.GetSession("auth"))
- if auth != u.Role_Admin {
- f.Write("n")
- return
- }
- codes := f.GetString("codes")
- userid := f.GetString("userid") //分配给谁
- codesarr := strings.Split(codes, ",") //被分配的爬虫
- user, _ := u.MgoEB.FindById("user", userid, nil)
- if user != nil && len(*user) > 0 {
- name := qu.ObjToString((*user)["s_name"])
- query := map[string]interface{}{
- "code": map[string]interface{}{
- "$in": codesarr,
- },
- }
- qu.Debug(query)
- set := map[string]interface{}{
- "$set": map[string]interface{}{
- "createuserid": userid,
- "createuser": name,
- "createuseremail": (*user)["s_email"],
- "modifyuser": name,
- "modifyuserid": userid,
- "platform": (*user)["s_platform"],
- },
- }
- //b := u.MgoE.Update("luaconfig", query, set, false, true)
- b := u.MgoEB.Update("luaconfig", query, set, false, true)
- if b {
- go editModify(codesarr, userid, name) //修改爬虫对应任务的维护人
- f.Write("y")
- } else {
- f.Write("n")
- }
- } else {
- f.Write("null")
- }
- }
- // 审核日志导出
- func (f *Front) AuditExport() {
- auth := qu.IntAll(f.GetSession("auth"))
- if auth > u.Role_Admin {
- exportnames := f.GetString("exportnames")
- exporttype := f.GetString("exporttype")
- starttime := f.GetString("starttime")
- endtime := f.GetString("endtime")
- qu.Debug(exportnames, exporttype, starttime, endtime)
- file, err := xlsx.OpenFile("爬虫审核记录.xlsx")
- if err != nil {
- f.Write("文件打开失败!")
- return
- }
- sheet := file.Sheets[0]
- exportLogs(sheet, exportnames, exporttype, starttime, endtime) //导出数据
- fname := fmt.Sprintf("爬虫审核打回记录%d.xlsx", time.Now().Unix())
- err = file.Save(fname)
- if err != nil {
- qu.Debug("Save Excel" + fname + "Error")
- f.ServeJson("导出失败")
- return
- }
- f.ResponseWriter.Header().Add("Content-Disposition", "attachment;filename=爬虫审核打回记录.xlsx")
- f.ServeFile(fname)
- go func(path string) { //删除
- time.Sleep(time.Second * 30)
- os.Remove(path)
- }(fname)
- } else {
- f.Write("您没有权限")
- }
- }
- func exportLogs(sheet *xlsx.Sheet, exportnames, exporttype, starttime, endtime string) {
- st, _ := time.ParseInLocation(qu.Date_Short_Layout, starttime, time.Local)
- et, _ := time.ParseInLocation(qu.Date_Short_Layout, endtime, time.Local)
- query := map[string]interface{}{
- "comeintime": map[string]interface{}{
- "$gte": st.Unix(),
- "$lt": et.Unix(),
- },
- }
- if exporttype == "repulse" { //打回
- query["types"] = "打回"
- } else if exporttype == "audit" { //审核
- query["types"] = "审核"
- }
- users := map[string]bool{}
- for _, user := range strings.Split(exportnames, ",") {
- users[user] = true
- }
- qu.Debug("audit log query:", query)
- sess := u.MgoEB.GetMgoConn()
- defer u.MgoEB.DestoryMongoConn(sess)
- lock := &sync.Mutex{}
- wg := &sync.WaitGroup{}
- ch := make(chan bool, 10)
- n := 0
- it := sess.DB(u.MgoEB.DbName).C("lua_logs_auditor").Find(&query).Iter()
- count, _ := sess.DB(u.MgoEB.DbName).C("lua_logs_auditor").Find(&query).Count()
- qu.Debug("count:", count)
- for tmp := map[string]interface{}{}; it.Next(tmp); n++ {
- ch <- true
- wg.Add(1)
- go func(tmp map[string]interface{}) {
- defer func() {
- <-ch
- wg.Done()
- }()
- id := mongodb.BsonIdToSId(tmp["_id"])
- auditor := qu.ObjToString(tmp["auditor"])
- code := qu.ObjToString(tmp["code"])
- comeintime := qu.Int64All(tmp["comeintime"])
- comeintime_z := qu.FormatDateByInt64(&comeintime, qu.Date_Full_Layout)
- event := qu.IntAll(tmp["event"])
- modifytime := qu.Int64All(tmp["modifytime"])
- modifytime_z := qu.FormatDateByInt64(&modifytime, qu.Date_Full_Layout)
- reason := qu.ObjToString(tmp["reason"])
- types := qu.ObjToString(tmp["types"])
- spideruser := qu.ObjToString(tmp["spideruser"])
- if !users[spideruser] {
- return
- }
- lock.Lock()
- row := sheet.AddRow()
- lock.Unlock()
- row.AddCell().SetValue(id)
- row.AddCell().SetValue(auditor)
- row.AddCell().SetValue(code)
- row.AddCell().SetValue(comeintime)
- row.AddCell().SetValue(comeintime_z)
- row.AddCell().SetValue(event)
- row.AddCell().SetValue(modifytime)
- row.AddCell().SetValue(modifytime_z)
- row.AddCell().SetValue(reason)
- row.AddCell().SetValue(spideruser)
- row.AddCell().SetValue(types)
- }(tmp)
- tmp = make(map[string]interface{})
- }
- }
- // 修改维护人
- func editModify(codesarr []string, userid, name string) {
- qu.Debug("修改爬虫对应任务的维护人:", name, codesarr)
- //修改modifyid和modify
- query := map[string]interface{}{
- "s_code": map[string]interface{}{
- "$in": codesarr,
- },
- "i_state": map[string]interface{}{
- "$nin": []int{4, 6},
- },
- }
- set := map[string]interface{}{
- "$set": map[string]interface{}{
- "s_modify": name,
- "s_modifyid": userid,
- },
- }
- b := u.MgoEB.Update("task", query, set, false, true)
- qu.Debug("任务重新分配成功", b)
- }
- func (f *Front) UpdatePendState() {
- success := false
- pendstate, _ := f.GetInteger("pendstate")
- code := f.GetString("code")
- stype := f.GetString("stype")
- qu.Debug(code, stype, pendstate)
- //success = u.MgoE.Update("luaconfig", map[string]interface{}{"code": code}, map[string]interface{}{"$set": map[string]interface{}{"pendstate": pendstate}}, false, false)
- success = u.MgoEB.Update("luaconfig", map[string]interface{}{"code": code}, map[string]interface{}{"$set": map[string]interface{}{"pendstate": pendstate, "pendtime": time.Now().Unix()}}, false, false)
- if success {
- if stype == "code" { //爬虫挂起时,关闭所有任务
- u.MgoEB.Update("task", map[string]interface{}{"s_code": code, "i_state": map[string]interface{}{"$ne": 6}}, map[string]interface{}{"$set": map[string]interface{}{"i_state": 6, "l_pendtime": time.Now().Unix()}}, false, true)
- } else if stype == "task" { //任务挂起任务时任务关闭
- id := f.GetString("id")
- u.MgoEB.UpdateById("task", id, map[string]interface{}{"$set": map[string]interface{}{"i_pendstate": pendstate, "i_state": 6, "l_pendtime": time.Now().Unix()}})
- }
- }
- f.ServeJson(map[string]interface{}{"success": success})
- }
- func (f *Front) UpdateESP() {
- val := f.GetString("val")
- w := f.GetString("w")
- id := f.GetString("id")
- code := f.GetString("c")
- query := map[string]interface{}{
- "_id": mongodb.StringTOBsonId(id),
- }
- set := map[string]interface{}{}
- update := map[string]interface{}{
- "$set": set,
- }
- //one, _ := u.MgoE.FindById("luaconfig", id, nil)
- one, _ := u.MgoEB.FindById("luaconfig", id, nil)
- if len(*one) == 0 {
- f.Write("n")
- return
- }
- if w == "infoformat" {
- infoformat, _ := f.GetInteger("val")
- set["infoformat"] = infoformat
- } else if w == "urgency" { //修改紧急度
- urgency, _ := f.GetInteger("val")
- set["urgency"] = urgency
- } else if w == "state" { //无效爬虫改为待完成
- tmpEvent := qu.IntAll((*one)["event"])
- if one != nil && len(*one) > 0 {
- if (*one)["incrementevent"] == nil && tmpEvent != 7000 { //除7000节点外没有incrementevent的要重新设置
- set["incrementevent"] = tmpEvent
- }
- }
- set["state"] = 0
- set["event"] = 7000
- set["spidertype"] = "history"
- go ModifyLogs_UpdateCodeState(code)
- //更新、新建任务
- //task, _ := u.MgoEB.FindOne("task", map[string]interface{}{"s_code": code, "i_state": map[string]interface{}{"$in": []int{0, 1, 2, 3, 5}}})
- //text := "指定追加描述:------------------------------\n无发布转待完成\n"
- //if len(*task) > 0 { //已有任务,更新
- // u.MgoEB.UpdateById("task", (*task)["_id"],
- // map[string]interface{}{
- // "$set": map[string]interface{}{
- // "i_state": 2,
- // "s_descript": qu.ObjToString((*task)["s_descript"]) + text,
- // "l_updatetime": time.Now().Unix(),
- // },
- // })
- //} else {
- // save := map[string]interface{}{
- // "s_channel": (*one)["channel"],
- // "i_event": (*one)["event"],
- // "i_frequencyerrtimes": 0,
- // "i_state": 2,
- // "s_descript": text,
- // "l_complete": time.Now().Unix() + int64(24*3600),
- // "s_urgency": "4",
- // "i_pendstate": 0,
- // "s_modifyid": (*one)["modifyuserid"],
- // "s_source": "人工",
- // "i_times": 0,
- // "l_comeintime": time.Now().Unix(),
- // "i_num": 0,
- // "s_modify": (*one)["modifyuser"],
- // "s_code": code,
- // "s_site": (*one)["site"],
- // "s_type": "0",
- // }
- // u.MgoEB.Save("task", save)
- //}
- } else if w == "platform" {
- set["platform"] = val
- set["comeintime"] = time.Now().Unix()
- if val != "golua平台" && val != "chrome" {
- b := u.MgoS.Update("spider_heart", map[string]interface{}{"code": code}, map[string]interface{}{"$set": map[string]interface{}{"del": true}}, false, true)
- qu.Debug("Del Heart:", b)
- b, err := spider.UpdateSpiderByCodeState(code, "6", qu.IntAll((*one)["event"])) //下架
- if b && err == nil {
- //历史节点下架为了避免线上运行爬虫待完成时改为7000采集历史,但是又转到其他平台,导致原线上运行节点爬虫并未下线,心跳异常
- if incrementevent := (*one)["incrementevent"]; incrementevent != nil {
- b, err = spider.UpdateSpiderByCodeState(code, "6", qu.IntAll(incrementevent))
- }
- //关闭任务
- u.MgoEB.Update("task", map[string]interface{}{"s_code": code, "i_state": map[string]interface{}{"$in": []int{0, 1, 2, 3, 5}}}, map[string]interface{}{"$set": map[string]interface{}{"i_state": 6}}, false, true)
- if b && err == nil {
- set["state"] = 0 //更新状态
- set["luauser"] = map[string]interface{}{
- "createuser": (*one)["createuser"],
- "createuserid": (*one)["createuserid"],
- "createuseremail": (*one)["createuseremail"],
- "modifyuser": (*one)["modifyuser"],
- "modifyuserid": (*one)["modifyuserid"],
- }
- } else {
- qu.Debug("历史节点下架失败")
- f.Write("n")
- return
- }
- } else {
- qu.Debug("增量节点下架失败")
- f.Write("n")
- return
- }
- //qu.Debug("下架:", upresult, code)
- //b, err := UpStateAndUpSpider(code, "", "", "", Sp_state_6) //线上爬虫下架
- //qu.Debug("爬虫下架成功:", b)
- //if !b || err != nil {
- // f.Write("n")
- // return
- //}
- }
- } else { //修改节点
- event, _ := strconv.Atoi(val)
- set["event"] = event
- set["incrementevent"] = event
- //state := f.GetString("s")
- //if state == "5" { //已上架状态改为下架
- code := f.GetString("c")
- set["state"] = 6
- b, err := UpStateAndUpSpider(code, "", "", "", Sp_state_6) //线上爬虫下架
- qu.Debug("爬虫下架成功:", b)
- if !b || err != nil {
- f.Write("n")
- return
- }
- //}
- }
- //if mgdb.Update("luaconfig", query, update, false, false) {
- if u.MgoEB.Update("luaconfig", query, update, false, false) {
- log.Println("Id:", id, " Update", w, val, "Success")
- f.Write("y")
- return
- } else {
- log.Println("Id:", id, " Update", w, val, "Failed")
- f.Write("n")
- return
- }
- f.Write("n")
- }
- func (f *Front) GetCity() {
- area := f.GetString("area")
- cityArr := []string{}
- cityArr = u.Province[area]
- f.ServeJson(cityArr)
- }
- func (f *Front) SpiderCopy() {
- defer qu.Catch()
- code := f.GetString("code")
- code = strings.TrimSpace(code)
- success := false
- text := ""
- encode := ""
- if code != "" {
- //lua := *mgdb.FindOne("luaconfig", map[string]interface{}{"code": code})
- lua, _ := u.MgoEB.FindOne("luaconfig", map[string]interface{}{"code": code})
- if len(*lua) > 0 {
- codeOld := qu.ObjToString((*lua)["code"])
- codeNew := codeOld + u.Bu
- //luaTmp := *mgdb.FindOne("luaconfig", map[string]interface{}{"code": codeNew})
- luaTmp, _ := u.MgoEB.FindOne("luaconfig", map[string]interface{}{"code": codeNew})
- if len(*luaTmp) > 0 {
- f.ServeJson(map[string]interface{}{"success": false, "text": "已存在补漏爬虫"})
- return
- }
- //u.MgoE.Update("luaconfig", map[string]interface{}{"code": codeOld}, map[string]interface{}{"$set": map[string]interface{}{"relatecode": codeNew}}, false, false)
- u.MgoEB.Update("luaconfig", map[string]interface{}{"code": codeOld}, map[string]interface{}{"$set": map[string]interface{}{"relatecode": codeNew}}, false, false)
- (*lua)["code"] = codeNew
- (*lua)["relatecode"] = codeOld
- if param_common := (*lua)["param_common"].([]interface{}); len(param_common) < 13 {
- f.ServeJson(map[string]interface{}{"success": false, "text": "脚本异常,请联系管理员"})
- return
- } else {
- param_common = param_common[:13]
- param_common[0] = codeNew //code
- param_common = append(param_common, true) //历史补漏
- param_common = append(param_common, false) //强制下载
- (*lua)["param_common"] = param_common
- }
- str_list := qu.ObjToString((*lua)["str_list"])
- (*lua)["str_list"] = strings.ReplaceAll(str_list, codeOld, codeNew)
- (*lua)["comeintime"] = time.Now().Unix()
- (*lua)["modifytime"] = time.Now().Unix()
- (*lua)["spidertype"] = "increment"
- (*lua)["event"] = 7000
- (*lua)["state"] = 6
- id := u.MgoEB.Save("luaconfig", (*lua))
- if id != "" {
- success = true
- encode = util.Se.Encode2Hex(codeNew)
- } else {
- text = "新爬虫保存失败"
- }
- } else {
- text = "未查询到复制原"
- }
- } else {
- text = "脚本不能为空"
- }
- f.ServeJson(map[string]interface{}{"success": success, "text": text, "encode": encode})
- }
- func (f *Front) SpiderSplitCopy() {
- defer qu.Catch()
- code := f.GetString("code")
- code = strings.TrimSpace(code)
- num, _ := f.GetInteger("num")
- text := ""
- lua, _ := u.MgoEB.FindOne("luaconfig", map[string]interface{}{"code": code})
- if len(*lua) > 0 {
- param_common := (*lua)["param_common"].([]interface{})
- str_list := qu.ObjToString((*lua)["str_list"])
- channelOld := qu.ObjToString((*lua)["channel"])
- copyNum := qu.IntAll((*lua)["copynum"]) //当前已经拆分了多少个爬虫
- delete(*lua, "relatecode") //删除关联爬虫字段
- for i := 1; i <= num; i++ {
- copyNum++
- var codeSuffix string
- if copyNum < 10 {
- codeSuffix = fmt.Sprintf("0%s", fmt.Sprint(copyNum))
- } else {
- codeSuffix = fmt.Sprint(copyNum)
- }
- (*lua)["code"] = code + "_" + codeSuffix
- (*lua)["channel"] = channelOld + "_" + codeSuffix
- (*lua)["str_list"] = strings.ReplaceAll(str_list, code, code+"_"+codeSuffix)
- (*lua)["comeintime"] = time.Now().Unix()
- (*lua)["modifytime"] = time.Now().Unix()
- (*lua)["state"] = 0
- param_common[0] = code + "_" + codeSuffix
- param_common[2] = channelOld + "_" + codeSuffix
- (*lua)["param_common"] = param_common
- u.MgoEB.Save("luaconfig", (*lua))
- }
- u.MgoEB.Update("luaconfig", map[string]interface{}{"code": code}, map[string]interface{}{"$set": map[string]interface{}{"copynum": copyNum}}, false, false)
- } else {
- text = "未查询到复制原"
- }
- f.ServeJson(map[string]interface{}{"text": text})
- }
- /*// 清理Redis
- func (f *Front) DelRedis() {
- hrefs := f.GetString("href")
- hrefsarr := strings.Split(hrefs, ",")
- auth := qu.IntAll(f.GetSession("auth"))
- err := []string{}
- if auth == u.Role_Admin { //权限控制
- if len(hrefsarr) > 0 {
- for k1, h := range hrefsarr {
- href := Reg.FindString(h)
- if href != "" {
- href = "url_repeat_" + href + "*"
- res := redis.GetKeysByPattern("title_repeat_judgement", href)
- if res != nil {
- for _, v := range res {
- hf := string(v.([]uint8))
- b := redis.Del("title_repeat_judgement", hf)
- if !b {
- err = append(err, "第"+strconv.Itoa(k1+1)+"个")
- }
- }
- }
- } else {
- err = append(err, "第"+strconv.Itoa(k1+1)+"个")
- }
- }
- }
- } else {
- err = append(err, "没有权限")
- }
- f.ServeJson(err)
- }*/
|