intimate/extractor/openrec_extractor/openrec_extractor.go

package main

import (
	"database/sql"
	"encoding/json"
	"intimate"
	"log"
	"os"
	"os/signal"
	"regexp"
	"strconv"
	"strings"
	"sync/atomic"
	"syscall"
	"time"

	"github.com/tidwall/gjson"
)

// OpenrecExtractor 提取方法
type OpenrecExtractor struct {
	user       *intimate.ExtractorSource
	userLive   *intimate.ExtractorSource
	supporters *intimate.ExtractorSource
}

func (oe *OpenrecExtractor) Execute() {

	var loop int32 = 1

	go func() {
		signalchan := make(chan os.Signal)
		signal.Notify(signalchan, syscall.SIGKILL, syscall.SIGQUIT, syscall.SIGTERM, syscall.SIGSTOP)
		log.Println("accept stop command:", <-signalchan)
		atomic.StoreInt32(&loop, 0)
	}()

	collect := intimate.NewExtractorStore()
	store := intimate.NewSourceStore("source_openrec")
	var lasterr error = nil

	for atomic.LoadInt32(&loop) > 0 {

		source, err := store.Pop(string(intimate.TTOpenrecRanking), 100)
		if err != nil {
			if err != lasterr {
				log.Println(err, lasterr)
				lasterr = err
			}
			time.Sleep(time.Second * 2)
			continue
		}

		source.SetOperator(int32(intimate.OperatorError))
		anchorId := source.GetSource().String

		ai := &intimate.AnchorInfo{}
		ai.SetAnchorId(anchorId)
		ai.SetPlatform(string(intimate.Popenrec))

		sdata := source.GetExt().([]byte)
		if gjson.ValidBytes(sdata) {
			result := gjson.ParseBytes(sdata)
			datamap := result.Map()

			oe.user = intimate.NewExtractorSource(datamap["user"])
			oe.user.CreateExtractor()

			oe.userLive = intimate.NewExtractorSource(datamap["user_live"])
			oe.userLive.CreateExtractor()

			oe.supporters = intimate.NewExtractorSource(datamap["supporters"])

			clog := &intimate.CollectLog{}

			log.Println(anchorId)

			oe.extractFollowers(clog)
			oe.extractAnchorName(ai)
			oe.extractViewsAndLiveStreaming(clog)
			oe.extractGiversAndGratuity(clog)
			oe.extractLive(clog)
			oe.extractTags(clog)

			ai.Set("UpdateTime", source.GetUpdateTime())

			LiveUrl := "https://www.openrec.tv/live/" + anchorId
			ai.Set("LiveUrl", sql.NullString{String: LiveUrl, Valid: true})

			Uid, err := collect.InsertAnchorInfo(ai)
			if err != nil {
				log.Println(err)
				source.SetErrorMsg(sql.NullString{String: err.Error(), Valid: true})
				store.UpdateOperator(source)
				return
			}

			clog.Set("Uid", Uid)
			clog.Set("Platform", string(intimate.Popenrec))
			clog.Set("AnchorId", anchorId)
			clog.Set("UpdateTime", source.GetUpdateTime())

			if err = collect.InsertCollectLog(clog); err != nil {
				source.SetErrorMsg(sql.NullString{String: err.Error(), Valid: true})
				store.UpdateOperator(source)
				return
			}

			source.SetOperator(int32(intimate.OperatorExtractorOK))
			store.UpdateOperator(source)
		} else {
			log.Println("data is not json:\n", string(sdata))
		}
	}
}

func (oe *OpenrecExtractor) extractFollowers(clog intimate.ISet) {
	extractor := oe.user.GetExtractor()
	xp, err := extractor.XPathResult("//p[@class='c-global__user__count__row__right js-userCountFollowers']/text()")
	if err != nil {
		log.Println(err)
	}
	if !xp.NodeIter().Next() {
		log.Println("不存在粉丝数")
	}

	followers := strings.ReplaceAll(xp.String(), ",", "")
	followersInt, err := strconv.ParseInt(followers, 10, 64)
	if err != nil {
		log.Println(err)
	}

	clog.Set("Followers", sql.NullInt64{Int64: followersInt, Valid: true})
}

func (oe *OpenrecExtractor) extractAnchorName(ai intimate.ISet) {
	extractor := oe.user.GetExtractor()
	xp, err := extractor.XPathResult("//p[@class='c-global__user__profile__list__name__text official-icon--after']/text()")
	if xp.NodeIter().Next() {
		anchorName := xp.String()
		ai.Set("AnchorName", anchorName)
	} else {
		log.Println(err)
	}
}

func (oe *OpenrecExtractor) extractViewsAndLiveStreaming(clog intimate.ISet) {
	extractor := oe.user.GetExtractor()
	// c-contents
	xp, err := extractor.XPathResult("//ul[@class='c-contents']//p[@class='c-thumbnailVideo__footer__liveCount']/text()")
	if err != nil {
		log.Println(err)
	}
	if xp.NodeIter().Next() {
		views := regexp.MustCompile(`[0-9,]+`).FindString(xp.String())
		views = strings.ReplaceAll(views, ",", "")
		viewsint, err := strconv.Atoi(views)
		if err != nil {
			log.Println(err)
		}

		clog.Set("Views", sql.NullInt64{Int64: int64(viewsint), Valid: true})
		clog.Set("IsLiveStreaming", int32(1))
	}
}

func (oe *OpenrecExtractor) extractGiversAndGratuity(clog intimate.ISet) {
	// extractor := oe.user.GetExtractor()
	giverjson := oe.supporters.GetSource()
	var givers []interface{}
	var gratuity int64 = 0

	for _, v := range giverjson.Array() {
		giverSource := gjson.Parse(v.String())
		for _, item := range giverSource.Get("data.items").Array() {
			givers = append(givers, item.Map())
			gratuity += item.Get("total_yells").Int()
		}
	}

	giversbytes, err := json.Marshal(givers)
	if err != nil {
		log.Println(err)
		clog.Set("ErrorMsg", sql.NullString{String: err.Error(), Valid: true})
	} else {
		clog.Set("Giver", giversbytes)
	}

	clog.Set("Gratuity", sql.NullInt64{Int64: gratuity, Valid: true})
}

func (oe *OpenrecExtractor) extractLive(clog intimate.ISet) {
	extractor := oe.userLive.GetExtractor()
	mathes := regexp.MustCompile("MovieTitle__Title[^>]+>(.{1,50})</h1>").FindStringSubmatch(oe.userLive.GetSource().Str)
	if len(mathes) == 2 {

		clog.Set("LiveTitle", sql.NullString{String: mathes[1], Valid: true})

		content, err := extractor.XPathResult("//meta[@itemprop='uploadDate']/@content")
		if err != nil {
			log.Println(err)
		}

		iter := content.NodeIter()
		if iter.Next() {
			tm, err := time.ParseInLocation("2006-01-02T15:04:05Z07:00", iter.Node().NodeValue(), time.Local)
			if err != nil {
				log.Println(err)
			}
			clog.Set("LiveStartTime", sql.NullTime{Time: tm.Local(), Valid: true})

			duration, err := extractor.XPathResult("//meta[@itemprop='duration']/@content")
			if err != nil {
				log.Println(err)
			}

			diter := duration.NodeIter()
			if diter.Next() {

				dt, err := intimate.ParseDuration(diter.Node().NodeValue())
				if err != nil {
					log.Println(err)
				}
				endtm := tm.Add(dt)
				clog.Set("LiveEndTime", sql.NullTime{Time: endtm.Local(), Valid: true})
			}
		}
	}
}

func (oe *OpenrecExtractor) extractTags(clog intimate.ISet) {
	var tags []string
	matheslist := regexp.MustCompile(`<[^>]+TagButton[^>]+>([^<]{1,100})<`).FindAllStringSubmatch(oe.userLive.GetSource().Str, -1)
	for _, m := range matheslist {
		tags = append(tags, m[1])
	}
	log.Println(tags)
	tagsBytes, err := json.Marshal(tags)
	if err != nil {
		log.Println(err)
	}

	clog.Set("Tags", tagsBytes)
}