Merge branch 'release/v0.6.0'

autostore.go

@@ -34,9 +34,10 @@ type Table struct {
 	name    string
 	setting interface{}
 
-	updatesql string
+	updatesql    string
-	selectsql string
+	selectsql    string
-	insertsql string
+	insertsql    string
+	duplicatesql string
 }
 
 func NewStore(uri string) *Store {
@@ -54,6 +55,7 @@ func (store *Store) Table(name string) *Table {
 	table.name = name
 
 	table.insertsql = `INSERT INTO ` + table.name + `(%s) values(%s)`
+	table.duplicatesql = `INSERT INTO ` + table.name + `(%s) values(%s) ON DUPLICATE KEY UPDATE %s`
 	table.updatesql = `UPDATE ` + table.name + ` SET %s WHERE %s = ?`
 	table.selectsql = `SELECT %s FROM ` + table.name + ` WHERE %s `
 	return table
@@ -226,6 +228,83 @@ func (t *Table) Insert(obj interface{}) error {
 	return err
 }
 
+// DUpdate ON DUPLICATE KEY UPDATE struct. Field对应的tag field字段
+type DUpdate struct {
+	Field string // selected 字段
+	Value interface{}
+}
+
+// InsertOrUpdate   nil 不插入. 不支持嵌套. 必须是Ptr类型
+func (t *Table) InsertOrUpdate(obj interface{}, updates ...DUpdate) error {
+	ov := reflect.ValueOf(obj).Elem()
+	ot := reflect.TypeOf(obj)
+
+	fieldsql := ""
+	argssql := ""
+
+	var SourceUpdate []*DUpdate
+	var OtherUpdate []*DUpdate
+	for _, u := range updates {
+		if u.Value == nil {
+			SourceUpdate = append(SourceUpdate, &u)
+		} else {
+			OtherUpdate = append(OtherUpdate, &u)
+		}
+	}
+
+	var args []interface{}
+	for i := 0; i < ov.NumField(); i++ {
+		field := ov.Field(i)
+		ftype := ot.Elem().Field(i)
+
+		if fname, ok := ftype.Tag.Lookup("field"); ok {
+			// if flag, ok := ftype.Tag.Lookup("uid"); ok {
+			// 	if flag == "auto" {
+			// 		continue
+			// 	}
+			// }
+
+			k := ftype.Type.Kind()
+			if k == reflect.Ptr || k == reflect.Interface {
+				if !field.IsNil() {
+					felem := field.Elem()
+					args = append(args, felem.Interface())
+					fieldsql += fname + ","
+					argssql += "?,"
+				}
+			} else {
+				args = append(args, field.Interface())
+				fieldsql += fname + ","
+				argssql += "?,"
+			}
+
+			for _, u := range SourceUpdate {
+				if u.Field == fname {
+					u.Value = args[len(args)-1]
+					break
+				}
+			}
+
+		}
+
+	}
+
+	var duplicateSet string = ""
+	for _, u := range SourceUpdate {
+		duplicateSet += u.Field + " = ?,"
+		args = append(args, u.Value)
+	}
+
+	for _, u := range OtherUpdate {
+		duplicateSet += u.Field + " = ?,"
+		args = append(args, u.Value)
+	}
+
+	ssql := fmt.Sprintf(t.duplicatesql, fieldsql[:len(fieldsql)-1], argssql[:len(argssql)-1], duplicateSet[:len(duplicateSet)-1])
+	_, err := t.store.db.Exec(ssql, args...)
+	return err
+}
+
 // InsertRetAutoID   nil 不插入. 不支持嵌套. 并返回auto uid
 func (t *Table) InsertRetAutoID(obj interface{}) (int64, error) {
 	ov := reflect.ValueOf(obj).Elem()

autostore_test.go

@@ -5,26 +5,24 @@ import (
 	"encoding/json"
 	"testing"
 	"time"
-
-	"github.com/davecgh/go-spew/spew"
 )
 
-func estAutoStore(t *testing.T) {
+func TestAutoStore(t *testing.T) {
 	uri := "root:@tcp(127.0.0.1:4000)/test?parseTime=true&loc=Local&charset=utf8mb4&collation=utf8mb4_unicode_ci"
 	store := NewStore(uri)
 
-	queue := store.Table("streamer").Queue(TSreamer{}, CondWhere{Condition: "operator = 0"})
+	// queue := store.Table("streamer").Queue(TSreamer{}, CondWhere{Condition: "operator = 0"})
-	re, _ := queue.Pop()
+	// re, _ := queue.Pop()
 
-	pstreamer := re.(*TSreamer)
+	// pstreamer := re.(*TSreamer)
-	m := make(map[string]interface{})
+	// m := make(map[string]interface{})
-	json.Unmarshal(pstreamer.Iface.([]byte), &m)
+	// json.Unmarshal(pstreamer.Iface.([]byte), &m)
-	spew.Println(re.(*TSreamer), m)
+	// spew.Println(re.(*TSreamer), m)
 
 	streamer := &TSreamer{}
 
-	streamer.Uid = 2
+	streamer.Uid = 1
-	streamer.UserID = &sql.NullString{String: "hehe", Valid: true}
+	streamer.UserID = &sql.NullString{String: "xixi", Valid: true}
 	streamer.Name = "streamer"
 	streamer.Operator = 0
 	streamer.Bit = 0b11
@@ -41,7 +39,7 @@ func estAutoStore(t *testing.T) {
 	now := time.Now()
 	streamer.UpdateTime = &now
 
-	err = store.Table("streamer").Insert(streamer)
+	err = store.Table("streamer").InsertOrUpdate(streamer, DUpdate{Field: "userid"})
 	if err != nil {
 		t.Error(err)
 	}

extractor/nimo_extractor/nimo_extractor.go

@@ -11,10 +11,10 @@ import (
 )
 
 // sstore  源存储实例, 为存储源数据的实现. 表格具体参考sql/intimate_source.sql
-var sstore *intimate.StoreSource = intimate.NewStoreSource(string(intimate.STNimo))
+// var sstore *intimate.StoreSource = intimate.NewStoreSource(string(intimate.STNimo))
 
-// estore 解析存储连接实例
+// // estore 解析存储连接实例
-var estore *intimate.StoreExtractor = intimate.NewStoreExtractor()
+// var estore *intimate.StoreExtractor = intimate.NewStoreExtractor()
 
 func main() {
 	Execute()
@@ -35,14 +35,18 @@ func Execute() {
 	waitfor := intimate.NewWaitFor(wd)
 	ps := intimate.NewPerfectShutdown()
 
+	queue := intimate.TStreamer.Queue(intimate.Streamer{}, intimate.ConditionDefault(intimate.PNimo))
+
 	for !ps.IsClose() {
-		streamer, err := estore.Pop(intimate.PNimo)
+		istreamer, err := queue.Pop()
 		if err != nil {
 			log.Println(err)
-			estore.UpdateError(streamer, err)
+			intimate.TStreamer.UpdateError(istreamer, err)
 			continue
 		}
 
+		streamer := istreamer.(*intimate.Streamer)
+
 		wd.Get(streamer.LiveUrl.String)
 		// wd.Get("https://www.nimo.tv/live/1253835677")
 
@@ -71,22 +75,25 @@ func Execute() {
 
 		clog := &intimate.CollectLog{}
 		clog.Platform = intimate.PNimo
-		clog.Followers = sql.NullInt64{Int64: li.Followers, Valid: true}
+		clog.Followers = &sql.NullInt64{Int64: li.Followers, Valid: true}
-		clog.Views = sql.NullInt64{Int64: li.Views, Valid: true}
+		clog.Views = &sql.NullInt64{Int64: li.Views, Valid: true}
-		clog.UpdateTime = utime
+		clog.UpdateTime = &utime
 		clog.StreamerUid = streamer.Uid
 
 		var sum int64 = 0
 		for _, v := range li.Gratuity {
 			sum += v
 		}
-		clog.Gratuity = sql.NullInt64{Int64: sum, Valid: true}
+		clog.Gratuity = &sql.NullInt64{Int64: sum, Valid: true}
 
-		cuid := estore.InsertClog(clog)
+		cuid, err := intimate.TClog.InsertRetAutoID(clog)
+		if err != nil {
+			panic(err)
+		}
 
-		streamer.Channel = sql.NullString{String: li.Channel, Valid: true}
+		streamer.Channel = &sql.NullString{String: li.Channel, Valid: true}
 		streamer.LatestLogUid = cuid
-		streamer.UpdateTime = utime
+		streamer.UpdateTime = &utime
 		streamer.Operator = 0
 
 		switch {
@@ -102,7 +109,11 @@ func Execute() {
 			streamer.UpdateInterval = 60
 		}
 
-		estore.Update(streamer, "update_interval", streamer.UpdateInterval, "operator", streamer.Operator, "channel", streamer.Channel, "latest_log_uid", streamer.LatestLogUid, "update_time", streamer.UpdateTime)
+		// estore.Update(streamer, "update_interval", streamer.UpdateInterval, "operator", streamer.Operator, "channel", streamer.Channel, "latest_log_uid", streamer.LatestLogUid, "update_time", streamer.UpdateTime)
+		err = intimate.TStreamer.Update(streamer)
+		if err != nil {
+			panic(err)
+		}
 
 		count++
 		if count >= countlimit {

extractor/twitch_extractor/tiwtch_extractor.go

@@ -3,6 +3,7 @@ package main
 import (
 	"database/sql"
 	"encoding/json"
+	"fmt"
 	"intimate"
 	"log"
 	"regexp"
@@ -12,134 +13,287 @@ import (
 	"github.com/tebeka/selenium"
 )
 
-// sstore  源存储实例, 为存储源数据的实现. 表格具体参考sql/intimate_source.sql
+// // sstore  源存储实例, 为存储源数据的实现. 表格具体参考sql/intimate_source.sql
-var sstore *intimate.StoreSource = intimate.NewStoreSource(string(intimate.STTwitch))
+// var sstore *intimate.StoreSource = intimate.NewStoreSource(string(intimate.STTwitch))
 
-// estore 解析存储连接实例
+// // estore 解析存储连接实例
-var estore *intimate.StoreExtractor = intimate.NewStoreExtractor()
+// var estore *intimate.StoreExtractor = intimate.NewStoreExtractor()
 
 func main() {
-	wd := intimate.GetChromeDriver(3030)
+	wd := intimate.GetChromeDriver(3040)
 	ps := intimate.NewPerfectShutdown()
 
+	queue := intimate.TStreamerList.Queue(intimate.StreamerList{}, intimate.ConditionDefault(intimate.Ptwitch))
 	var count = 0
 	var countlimt = 200
 
-	var lasterr error = nil
+	// var lasterr error = nil
 	// var err error
-
 	for !ps.IsClose() {
-		streamer, err := estore.Pop(intimate.Ptwitch)
+
-		if streamer == nil || err != nil {
+		// sourceChannel, err := sstore.Pop(intimate.TTwitchChannel)
-			if err != lasterr {
+		isl, err := queue.Pop()
-				log.Println(err, lasterr)
+		if err != nil {
-				lasterr = err
+			panic(err)
-			}
-			time.Sleep(time.Second * 2)
-			continue
 		}
+		streamerlist := isl.(*intimate.StreamerList)
 
-		var updateUrl map[string]string
+		weburl := streamerlist.Url + "?sort=VIEWER_COUNT"
-		json.Unmarshal(streamer.UpdateUrl.([]byte), &updateUrl)
+		err = wd.Get(weburl)
-		liveUrl := updateUrl["live"]
-		liveUrl = strings.Replace(liveUrl, "/watchparty", "", -1)
-		log.Println(liveUrl)
-
-		// err = wd.Get("https://www.twitch.tv/zoe_0601" + "/about")
-		err = wd.Get(liveUrl + "/about")
 		if err != nil {
 			log.Println(err)
-			estore.UpdateError(streamer, err)
+			// sstore.UpdateError(sourceChannel, err)
-			time.Sleep(time.Second * 5)
+			intimate.TStreamerList.UpdateError(streamerlist, err)
+			time.Sleep(time.Second * 10)
 			continue
 		}
 
-		streamer.LiveUrl = sql.NullString{String: liveUrl, Valid: true}
+		wd.WaitWithTimeout(func(wd selenium.WebDriver) (bool, error) {
-		clog := &intimate.CollectLog{}
+			_, err := wd.FindElement(selenium.ByXPATH, "(//div/p[@class=''])[last()]")
-		clog.UserId = streamer.UserId
+			if err != nil {
-		clog.Gratuity = sql.NullInt64{Int64: 0, Valid: false}
+				return false, err
-
-		time.Sleep(time.Millisecond * 500)
-		err = extractUserName(wd, streamer)
-		if err != nil {
-			_, err = wd.FindElement(selenium.ByXPATH, "//a[@data-a-target='browse-channels-button']")
-			if err == nil {
-				log.Println(streamer.UserId, "may be cancell")
-				streamer.Operator = 5
-				streamer.UpdateTime = sql.NullTime{Time: time.Now(), Valid: true}
-				estore.UpdateStreamer(streamer)
 			}
-			continue
+			return true, nil
-		}
+		}, time.Second*10)
-		err = extractFollowers(wd, clog)
-		if err != nil {
-			continue
-		}
 
-		err = extractViews(wd, clog) // views + tags + gratuity
+		btn, err := wd.FindElement(selenium.ByXPATH, "//button[@data-a-target='browse-sort-menu']")
 		if err != nil {
-			// 不直播时提取礼物 gratuity
+			log.Println(err)
-			wd.WaitWithTimeout(func(web selenium.WebDriver) (bool, error) {
+			continue
-				channelchat, err := wd.FindElement(selenium.ByXPATH, `//a[@data-a-target="channel-home-tab-Chat"]`)
+		}
-				btn, _ := web.FindElement(selenium.ByXPATH, `//button[@data-test-selector="expand-grabber"]`)
+		btn.Click()
-				if (err == nil && channelchat != nil) || btn != nil {
+
-					if channelchat != nil {
+		var elements []selenium.WebElement
-						channelchat.Click()
+		var liveurls = 0
-					}
+		var delayerror = 2
-					time.Sleep(time.Second)
+		for i := 0; i < 200 && !ps.IsClose(); i++ {
-					extractGratuity(wd, clog)
+			elements, err = wd.FindElements(selenium.ByXPATH, "(//div/p[@class=''])[last()]")
-					return true, nil
+			if err != nil {
+				log.Println(err)
+				break
+			}
+			time.Sleep(time.Millisecond * 200)
+			wd.KeyDown(selenium.EndKey)
+			time.Sleep(time.Millisecond * 200)
+			wd.KeyUp(selenium.EndKey)
+			time.Sleep(time.Millisecond * 2000)
+			if len(elements) == liveurls {
+				delayerror--
+				if delayerror <= 0 {
+					break
 				}
-				return false, nil
+			} else {
-
+				delayerror = 2
-			}, time.Second*4)
+			}
+			liveurls = len(elements)
+		}
+		articles, err := wd.FindElements(selenium.ByXPATH, "//article")
+		if err != nil {
+			log.Println(err)
+			continue
 		}
 
-		streamer.Platform = intimate.Ptwitch
+		var streamers []*intimate.Streamer
-		clog.Platform = streamer.Platform
+		for _, article := range articles {
-		clog.UpdateTime = sql.NullTime{Time: time.Now(), Valid: true}
-		lastClogId := estore.InsertClog(clog)
 
-		streamer.Operator = 10
+			e, err := article.FindElement(selenium.ByXPATH, ".//a[@data-a-target='preview-card-title-link' and @href]")
-		streamer.LatestLogUid = lastClogId
+			if err != nil {
-		if clog.Tags != nil {
+				log.Println(err)
-			streamer.Tags = clog.Tags
+				continue
+			}
+
+			href, err := e.GetAttribute("href")
+			if err != nil {
+				log.Println(err)
+				continue
+			}
+
+			btns, err := article.FindElements(selenium.ByXPATH, ".//div[@class='tw-full-width tw-inline-block']//button")
+			if err != nil {
+				log.Println(err)
+				continue
+			}
+
+			var tags []string
+			for _, btn := range btns {
+				tag, err := btn.GetAttribute("data-a-target")
+				if err == nil {
+					tags = append(tags, tag)
+				}
+			}
+
+			streamer := &intimate.Streamer{}
+			matches := regexp.MustCompile(`https://www.twitch.tv/(\w+)`).FindStringSubmatch(href)
+			if len(matches) == 2 {
+				mc := matches[1]
+				streamer.UserId = &mc
+			} else {
+				log.Println(href)
+				continue
+			}
+
+			jtags, err := json.Marshal(tags)
+			if err != nil {
+				log.Println(err)
+			} else {
+				streamer.Tags = jtags
+			}
+
+			streamer.Platform = intimate.Ptwitch
+			streamer.LiveUrl = &sql.NullString{String: href, Valid: true}
+			streamer.Operator = 0
+
+			streamers = append(streamers, streamer)
+
+			// if estore.InsertStreamer(streamer) {
+			// 	// log.Println("streamer update tags", streamer.Uid, tags)
+			// 	if streamer.Tags != nil {
+			// 		estore.Update(streamer, "Tags", streamer.Tags)
+			// 	}
+			// }
 		}
 
-		switch fl := clog.Followers.Int64; {
+		for _, streamer := range streamers {
-		case fl > 100000:
+			Extractor(wd, streamer)
-			streamer.UpdateInterval = 120
+			if err = intimate.TStreamer.InsertOrUpdate(streamer,
-		case fl > 10000:
+				intimate.DUpdate{Field: "tags"},
-			streamer.UpdateInterval = 240 * 2
+				intimate.DUpdate{Field: "update_time"},
-		case fl > 1000:
+			); err != nil {
-			streamer.UpdateInterval = 360 * 2
+				log.Println(err)
-		case fl > 100:
+			}
-			streamer.UpdateInterval = 720 * 2
+		}
-		case fl > 0:
+
-			streamer.UpdateInterval = 1440 * 4
+		log.Println("streamer find", len(articles))
+		if len(articles) == 0 {
+			intimate.TStreamerList.UpdateError(streamerlist, fmt.Errorf(""))
 		}
 
-		streamer.UpdateTime = clog.UpdateTime
-		estore.UpdateStreamer(streamer)
 		count++
 		if count >= countlimt {
 			count = 0
-			// wd.Quit()
+			wd = intimate.GetChromeDriver(3031)
-			wd = intimate.GetChromeDriver(3030)
 		}
+
 	}
 
 	wd.Close()
 	wd.Quit()
 }
 
+func Extractor(wd selenium.WebDriver, streamer *intimate.Streamer) {
+	// streamer, err := estore.Pop(intimate.Ptwitch)
+	// if streamer == nil || err != nil {
+	// 	if err != lasterr {
+	// 		log.Println(err, lasterr)
+	// 		lasterr = err
+	// 	}
+	// 	time.Sleep(time.Second * 2)
+	// 	continue
+	// }
+
+	// var updateUrl map[string]string
+	// json.Unmarshal(streamer.UpdateUrl.([]byte), &updateUrl)
+	liveUrl := streamer.LiveUrl.String
+
+	liveUrl = strings.Replace(liveUrl, "/watchparty", "", -1)
+	log.Println(liveUrl)
+
+	// err = wd.Get("https://www.twitch.tv/zoe_0601" + "/about")
+	err := wd.Get(liveUrl + "/about")
+	if err != nil {
+		log.Println(err)
+		intimate.TStreamer.UpdateError(streamer, err)
+		time.Sleep(time.Second * 5)
+		return
+	}
+
+	streamer.LiveUrl = &sql.NullString{String: liveUrl, Valid: true}
+	clog := &intimate.CollectLog{}
+	clog.UserId = *streamer.UserId
+	clog.Gratuity = &sql.NullInt64{Int64: 0, Valid: false}
+
+	time.Sleep(time.Millisecond * 500)
+	err = extractUserName(wd, streamer)
+	if err != nil {
+		_, err = wd.FindElement(selenium.ByXPATH, "//a[@data-a-target='browse-channels-button']")
+		if err == nil {
+			log.Println(streamer.UserId, "may be cancell")
+			streamer.Operator = 5
+			streamer.UpdateTime = &sql.NullTime{Time: time.Now(), Valid: true}
+			intimate.TStreamer.UpdateError(streamer, fmt.Errorf(""))
+		}
+		return
+	}
+	err = extractFollowers(wd, clog)
+	if err != nil {
+		return
+	}
+
+	err = extractViews(wd, clog) // views + tags + gratuity
+	if err != nil {
+		// 不直播时提取礼物 gratuity
+		wd.WaitWithTimeout(func(web selenium.WebDriver) (bool, error) {
+			channelchat, err := wd.FindElement(selenium.ByXPATH, `//a[@data-a-target="channel-home-tab-Chat"]`)
+			btn, _ := web.FindElement(selenium.ByXPATH, `//button[@data-test-selector="expand-grabber"]`)
+			if (err == nil && channelchat != nil) || btn != nil {
+				if channelchat != nil {
+					channelchat.Click()
+				}
+				time.Sleep(time.Second)
+				extractGratuity(wd, clog)
+				return true, nil
+			}
+			return false, nil
+
+		}, time.Second*4)
+	}
+
+	streamer.Platform = intimate.Ptwitch
+	clog.Platform = streamer.Platform
+	clog.UpdateTime = &sql.NullTime{Time: time.Now(), Valid: true}
+	// clog.StreamerUid = streamer.Uid
+	lastClogId, err := intimate.TClog.InsertRetAutoID(clog)
+	if err != nil {
+		log.Println(err)
+		return
+	}
+
+	streamer.Operator = 10
+	streamer.LatestLogUid = lastClogId
+	if clog.Tags != nil {
+		streamer.Tags = clog.Tags
+	}
+
+	switch fl := clog.Followers.Int64; {
+	case fl > 100000:
+		streamer.UpdateInterval = 120
+	case fl > 10000:
+		streamer.UpdateInterval = 240 * 2
+	case fl > 1000:
+		streamer.UpdateInterval = 360 * 2
+	case fl > 100:
+		streamer.UpdateInterval = 720 * 2
+	case fl > 0:
+		streamer.UpdateInterval = 1440 * 4
+	}
+
+	streamer.UpdateTime = clog.UpdateTime
+	// intimate.TStreamer.InsertOrUpdate(streamer)
+	// count++
+	// if count >= countlimt {
+	// 	count = 0
+	// 	// wd.Quit()
+	// 	wd = intimate.GetChromeDriver(3030)
+	// }
+}
+
 func extractUserName(wd selenium.WebDriver, streamer *intimate.Streamer) error {
 	return wd.WaitWithTimeout(func(web selenium.WebDriver) (bool, error) {
 		label, err := web.FindElement(selenium.ByXPATH, "//a[@class='tw-interactive']//h1")
 		if err == nil {
 			if ltxt, err := label.Text(); err == nil && ltxt != "" {
 				// log.Println("label:", ltxt)
-				streamer.UserName = sql.NullString{String: ltxt, Valid: true}
+				streamer.UserName = &sql.NullString{String: ltxt, Valid: true}
 				return true, nil
 			}
 		}
@@ -159,7 +313,7 @@ func extractFollowers(wd selenium.WebDriver, clog *intimate.CollectLog) error {
 		}
 		followers = regexp.MustCompile(`[\d,]+`).FindString(followers)
 		fint, _ := intimate.ParseNumber(followers)
-		clog.Followers = sql.NullInt64{Int64: int64(fint), Valid: true}
+		clog.Followers = &sql.NullInt64{Int64: int64(fint), Valid: true}
 		// log.Println("followers: ", followers, fint)
 		return true, nil
 	}, 4*time.Second)
@@ -172,7 +326,7 @@ func extractViews(wd selenium.WebDriver, clog *intimate.CollectLog) error {
 			if txt, err := views.Text(); err == nil {
 
 				vint, _ := intimate.ParseNumber(txt)
-				clog.Views = sql.NullInt64{Int64: vint, Valid: true}
+				clog.Views = &sql.NullInt64{Int64: vint, Valid: true}
 				// log.Println("views:", txt)
 				views.Click()
 
@@ -192,7 +346,7 @@ func extractTitle(wd selenium.WebDriver, clog *intimate.CollectLog) error {
 		title, err := web.FindElement(selenium.ByXPATH, `//h2[@data-a-target='stream-title']`)
 		if err == nil {
 			if txt, err := title.Text(); err == nil {
-				clog.LiveTitle = sql.NullString{String: txt, Valid: true}
+				clog.LiveTitle = &sql.NullString{String: txt, Valid: true}
 				return true, nil
 			}
 		}
@@ -244,7 +398,7 @@ func extractGratuity(wd selenium.WebDriver, clog *intimate.CollectLog) error {
 						log.Println(err)
 					}
 				}
-				clog.Gratuity = sql.NullInt64{Int64: gratuity, Valid: true}
+				clog.Gratuity = &sql.NullInt64{Int64: gratuity, Valid: true}
 			}
 			return true, nil
 		}

extractor_field.go

@@ -9,9 +9,9 @@ type GetSet struct {
 }
 
 type StreamerList struct {
-	UrlHash  string `field:"urlhash" `  //
+	UrlHash  string `field:"urlhash" uid:"true"` //
-	Platform string `field:"platform" ` //
+	Platform string `field:"platform" `          //
-	Url      string `field:"url" `      //
+	Url      string `field:"url" `               //
 
 	Label *sql.NullString `field:"label" ` //
 

supervisor_conf/openrec_task2.conf

@@ -1,12 +0,0 @@
-[supervisord]
-nodaemon=true
-
-[program:openrec_source]
-directory = MYPATH/bin/openrec_task2/
-command= MYPATH/bin/openrec_task2/openrec_task2
-process_name=%(program_name)s_%(process_num)02d              ;多进程名称
-numprocs=4                                                   ;启动多个进程
-autorestart=true
-stderr_logfile=MYPATH/bin/openrec_task2/log
-stderr_logfile_maxbytes=0
-stopsignal=QUIT

supervisor_conf/twitch_extractor.conf

@@ -1,5 +1,5 @@
 [supervisord]
-nodaemon=false
+nodaemon=true
 
 [program:twitch_extractor]
 environment=DISPLAY=":99"

supervisor_conf/twitch_extractor_p2.conf

@@ -1,13 +0,0 @@
-[supervisord]
-nodaemon=false
-
-[program:twitch_extractor_p2]
-environment=DISPLAY=":99",pac_proxy=http://localhost:1090/pac1
-directory = MYPATH/bin/twitch_extractor
-command= MYPATH/bin/twitch_extractor/twitch_extractor 
-process_name=%(program_name)s_%(process_num)02d              ;多进程名称
-numprocs=2                                                  ;启动多个进程
-autorestart=true
-stderr_logfile=MYPATH/bin/twitch_extractor/log
-stderr_logfile_maxbytes=0
-stopsignal=QUIT

supervisor_conf/twitch_task1.conf

@@ -0,0 +1,13 @@
+[supervisord]
+nodaemon=false
+
+[program:twitch_task1]
+environment=DISPLAY=":99"
+directory = MYPATH/bin/twitch_task1
+command= MYPATH/bin/twitch_task1/twitch_task1 
+# process_name=%(program_name)s_%(process_num)02d              ;多进程名称
+# numprocs=1                                                   ;启动多个进程
+autorestart=true
+stderr_logfile=MYPATH/bin/twitch_task1/log
+stderr_logfile_maxbytes=0
+stopsignal=QUIT

supervisor_conf/twitch_task2.conf

@@ -1,13 +0,0 @@
-[supervisord]
-nodaemon=false
-
-[program:twitch_task2]
-environment=DISPLAY=":99"
-directory = MYPATH/bin/twitch_task2
-command= MYPATH/bin/twitch_task2/twitch_task2 
-process_name=%(program_name)s_%(process_num)02d              ;多进程名称
-numprocs=6                                                   ;启动多个进程
-autorestart=true
-stderr_logfile=MYPATH/bin/twitch_task2/log
-stderr_logfile_maxbytes=0
-stopsignal=QUIT

tasks/nimo/nimo_task1/nimo_task1.go

@@ -12,7 +12,7 @@ import (
 )
 
 // estore 解析存储连接实例
-var estore *intimate.StoreExtractor = intimate.NewStoreExtractor()
+// var estore *intimate.StoreExtractor = intimate.NewStoreExtractor()
 
 // Execute 执行
 func Execute() {
@@ -95,7 +95,10 @@ func Execute() {
 				}
 
 				streamer.UpdateInterval = 120
-				estore.InsertStreamer(streamer)
+				err = intimate.TStreamer.Insert(streamer)
+				if err != nil {
+					panic(err)
+				}
 
 			} else {
 				log.Println("userid is null.", room.String())

tasks/twitcasting/twitcasting_task1/twitcasting.go

@@ -70,11 +70,11 @@ func Execute() {
 			if ok := queuedict[wurl]; !ok {
 				log.Println(wurl)
 				sl := &intimate.StreamerList{}
-				sl.Platform = intimate.Ptwitcasting
+				sl.Platform = string(intimate.Ptwitcasting)
 				sl.Url = wurl
 				sl.Operator = 0
 				sl.UpdateInterval = 120
-				sl.UpdateTime = time.Now()
+				sl.UpdateTime = &sql.NullTime{Time: time.Now(), Valid: true}
 				sl.UrlHash = intimate.GetUrlHash(sl.Url)
 
 				intimate.TStreamerList.Insert(sl)
@@ -104,11 +104,11 @@ func Execute() {
 				sp.TagUrl[i] = wurl
 				if ok := queuedict[wurl]; !ok {
 					sl := &intimate.StreamerList{}
-					sl.Platform = intimate.Ptwitcasting
+					sl.Platform = string(intimate.Ptwitcasting)
 					sl.Url = wurl
 					sl.Operator = 0
 					sl.UpdateInterval = 120
-					sl.UpdateTime = time.Now()
+					sl.UpdateTime = &sql.NullTime{Time: time.Now(), Valid: true}
 					sl.UrlHash = intimate.GetUrlHash(sl.Url)
 					intimate.TStreamerList.Insert(sl)
 

tasks/twitch/twitch_task2/.gitignore

@@ -1,2 +0,0 @@
-twitch_task2
-log

tasks/twitch/twitch_task2/main.go

@@ -1,6 +0,0 @@
-package main
-
-func main() {
-
-	Execute()
-}

tasks/twitch/twitch_task2/task_twitch.go

@@ -1,175 +0,0 @@
-package main
-
-import (
-	"database/sql"
-	"encoding/json"
-	"intimate"
-	"log"
-	"regexp"
-	"time"
-
-	"github.com/tebeka/selenium"
-)
-
-// // sstore  源存储实例, 为存储源数据的实现. 表格具体参考sql/intimate_source.sql
-// var sstore *intimate.StoreSource = intimate.NewStoreSource(string(intimate.STTwitch))
-
-// // estore 解析存储连接实例
-// var estore *intimate.StoreExtractor = intimate.NewStoreExtractor()
-
-// 获取类型的所有频道链接
-
-// Execute 执行任务
-func Execute() {
-	// DELETE FROM source_twitch WHERE uid NOT IN (SELECT MAX(s.uid) FROM (SELECT uid, source FROM source_twitch ) s GROUP BY s.source) ;
-	//article//a[@data-a-target='preview-card-title-link']
-
-	wd := intimate.GetChromeDriver(3030)
-	defer wd.Quit()
-	ps := intimate.NewPerfectShutdown()
-	counter := intimate.NewCounter()
-	counter.SetMaxLimit(100)
-	counter.SetMaxToDo(func(olist ...interface{}) error {
-		owd := olist[0].(*selenium.WebDriver)
-		if err := (*owd).Quit(); err != nil {
-			log.Println(err)
-		}
-		*owd = intimate.GetChromeDriver(3030)
-		return nil
-	}, &wd)
-
-	for !ps.IsClose() {
-
-		var err error
-		sourceChannel, err := sstore.Pop(intimate.TTwitchChannel)
-		if err != nil {
-			panic(err)
-		}
-
-		weburl := sourceChannel.Source.String + "?sort=VIEWER_COUNT"
-		err = wd.Get(weburl)
-		if err != nil {
-			log.Println(err)
-			sstore.UpdateError(sourceChannel, err)
-			time.Sleep(time.Second * 10)
-			continue
-		}
-
-		wd.WaitWithTimeout(func(wd selenium.WebDriver) (bool, error) {
-			_, err := wd.FindElement(selenium.ByXPATH, "(//div/p[@class=''])[last()]")
-			if err != nil {
-				return false, err
-			}
-			return true, nil
-		}, time.Second*10)
-
-		btn, err := wd.FindElement(selenium.ByXPATH, "//button[@data-a-target='browse-sort-menu']")
-		if err != nil {
-			log.Println(err)
-			continue
-		}
-		btn.Click()
-
-		var elements []selenium.WebElement
-		var liveurls = 0
-		var delayerror = 2
-		for i := 0; i < 200 && !ps.IsClose(); i++ {
-			elements, err = wd.FindElements(selenium.ByXPATH, "(//div/p[@class=''])[last()]")
-			if err != nil {
-				log.Println(err)
-				break
-			}
-			time.Sleep(time.Millisecond * 200)
-			wd.KeyDown(selenium.EndKey)
-			time.Sleep(time.Millisecond * 200)
-			wd.KeyUp(selenium.EndKey)
-			time.Sleep(time.Millisecond * 2000)
-			if len(elements) == liveurls {
-				delayerror--
-				if delayerror <= 0 {
-					break
-				}
-			} else {
-				delayerror = 2
-			}
-			liveurls = len(elements)
-		}
-		articles, err := wd.FindElements(selenium.ByXPATH, "//article")
-		if err != nil {
-			log.Println(err)
-			continue
-		}
-
-		for _, article := range articles {
-
-			e, err := article.FindElement(selenium.ByXPATH, ".//a[@data-a-target='preview-card-title-link' and @href]")
-			if err != nil {
-				log.Println(err)
-				continue
-			}
-
-			href, err := e.GetAttribute("href")
-			if err != nil {
-				log.Println(err)
-				continue
-			}
-
-			btns, err := article.FindElements(selenium.ByXPATH, ".//div[@class='tw-full-width tw-inline-block']//button")
-			if err != nil {
-				log.Println(err)
-				continue
-			}
-
-			var tags []string
-			for _, btn := range btns {
-				tag, err := btn.GetAttribute("data-a-target")
-				if err == nil {
-					tags = append(tags, tag)
-				}
-			}
-
-			streamer := &intimate.Streamer{}
-
-			matches := regexp.MustCompile(`https://www.twitch.tv/(\w+)`).FindStringSubmatch(href)
-			if len(matches) == 2 {
-				mc := matches[1]
-				streamer.UserId = &mc
-			} else {
-				log.Println(href)
-				continue
-			}
-
-			jtags, err := json.Marshal(tags)
-			if err != nil {
-				log.Println(err)
-			} else {
-				streamer.Tags = jtags
-			}
-
-			streamer.Platform = intimate.Ptwitch
-
-			updateUrl := make(map[string]string)
-			updateUrl["live"] = href
-			streamer.LiveUrl = &sql.NullString{String: href, Valid: true}
-			data, err := json.Marshal(updateUrl)
-			if err != nil {
-				log.Println(err)
-				continue
-			}
-			streamer.UpdateUrl = data
-			streamer.Operator = 0
-			if estore.InsertStreamer(streamer) {
-				// log.Println("streamer update tags", streamer.Uid, tags)
-				if streamer.Tags != nil {
-					estore.Update(streamer, "Tags", streamer.Tags)
-				}
-			}
-		}
-		log.Println("streamer find", len(articles))
-		if len(articles) == 0 {
-			sourceChannel.Operator = 5
-			sstore.UpdateOperator(sourceChannel)
-		}
-		counter.AddWithReset(1)
-	}
-}

tasks/twitch/twitch_task2/task_twitch_test.go

@@ -1,7 +0,0 @@
-package main
-
-import "testing"
-
-func TestMain(t *testing.T) {
-	main()
-}