完成twitch的期望功能

This commit is contained in:
eson
2020-07-31 18:04:10 +08:00
parent ac1ab81676
commit 0c54cbf8d8
13 changed files with 334 additions and 100 deletions

View File

@@ -25,7 +25,7 @@ type ChannelLink struct {
func (cl *ChannelLink) Execute() {
var err error
wd := intimate.GetChromeDriver(3030)
defer wd.Close()
ps := intimate.NewPerfectShutdown()
weburl := "https://www.twitch.tv/directory?sort=VIEWER_COUNT"
err = wd.Get(weburl)
@@ -40,7 +40,7 @@ func (cl *ChannelLink) Execute() {
}
return len(elements) > 0, nil
}
wd.WaitWithTimeout(cardCondition, time.Second*30)
wd.WaitWithTimeout(cardCondition, time.Second*15)
time.Sleep(time.Second)
e, err := wd.FindElement(selenium.ByXPATH, "//button[@data-a-target='browse-sort-menu']")
@@ -68,15 +68,22 @@ func (cl *ChannelLink) Execute() {
delayerror = 5
}
for _, card := range cards {
href, err := card.GetAttribute("href")
if err != nil {
log.Println(err)
} else {
hrefs[href] = true
for ii := 0; ii < 10; ii++ {
for _, card := range cards {
href, err := card.GetAttribute("href")
if err != nil {
log.Println(href, err)
continue
} else {
hrefs[href] = true
}
}
break
}
samecount = len(cards)
if ps.IsClose() {
break
}
if len(cards) > 10 {
log.Println(len(cards))
@@ -88,7 +95,6 @@ func (cl *ChannelLink) Execute() {
time.Sleep(time.Millisecond * 200)
wd.KeyUp(selenium.EndKey)
time.Sleep(time.Millisecond * 2500)
}
for href := range hrefs {

View File

@@ -5,11 +5,7 @@ import (
"encoding/json"
"intimate"
"log"
"os"
"os/signal"
"regexp"
"sync/atomic"
"syscall"
"time"
"github.com/tebeka/selenium"
@@ -33,19 +29,18 @@ func (cl *UserList) Execute() {
//article//a[@data-a-target='preview-card-title-link']
wd := intimate.GetChromeDriver(3030)
defer wd.Close()
ps := intimate.NewPerfectShutdown()
counter := intimate.NewCounter()
counter.SetMaxLimit(100)
counter.SetMaxToDo(func(olist ...interface{}) error {
owd := olist[0].(*selenium.WebDriver)
(*owd).Close()
(*owd).Quit()
*owd = intimate.GetChromeDriver(3030)
return nil
}, &wd)
var loop int32 = 1
var count = 0
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)
}()
for atomic.LoadInt32(&loop) > 0 {
for !ps.IsClose() {
var err error
sourceChannel, err := sstore.Pop(intimate.TTwitchChannel)
@@ -56,7 +51,10 @@ func (cl *UserList) Execute() {
weburl := sourceChannel.Source.String + "?sort=VIEWER_COUNT"
err = wd.Get(weburl)
if err != nil {
panic(err)
log.Println(err)
sstore.UpdateError(sourceChannel, err)
time.Sleep(time.Second * 10)
continue
}
wd.WaitWithTimeout(func(wd selenium.WebDriver) (bool, error) {
@@ -77,16 +75,17 @@ func (cl *UserList) Execute() {
var elements []selenium.WebElement
var liveurls = 0
var delayerror = 2
for i := 0; i < 200 && atomic.LoadInt32(&loop) > 0; i++ {
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 * 500)
time.Sleep(time.Millisecond * 200)
wd.KeyDown(selenium.EndKey)
time.Sleep(time.Millisecond * 200)
wd.KeyUp(selenium.EndKey)
time.Sleep(time.Millisecond * 1500)
time.Sleep(time.Millisecond * 2000)
if len(elements) == liveurls {
delayerror--
if delayerror <= 0 {
@@ -97,34 +96,62 @@ func (cl *UserList) Execute() {
}
liveurls = len(elements)
}
elements, err = wd.FindElements(selenium.ByXPATH, "//article//a[@data-a-target='preview-card-title-link' and @href]")
articles, err := wd.FindElements(selenium.ByXPATH, "//article")
if err != nil {
log.Println(err)
continue
}
for _, e := range elements {
for _, article := range articles {
attr, err := e.GetAttribute("href")
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(attr)
matches := regexp.MustCompile(`https://www.twitch.tv/(\w+)`).FindStringSubmatch(href)
if len(matches) == 2 {
streamer.UserId = matches[1]
} else {
log.Println(attr)
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"] = attr
streamer.LiveUrl = sql.NullString{String: attr, Valid: true}
updateUrl["live"] = href
streamer.LiveUrl = sql.NullString{String: href, Valid: true}
data, err := json.Marshal(updateUrl)
if err != nil {
log.Println(err)
@@ -132,15 +159,19 @@ func (cl *UserList) Execute() {
}
streamer.UpdateUrl = data
streamer.Operator = 0
estore.InsertStreamer(streamer)
if estore.InsertStreamer(streamer) {
// log.Println("streamer update tags", streamer.Uid, tags)
estore.Update(streamer, "Tags", streamer.Tags)
}
}
log.Println("streamer insert", len(elements))
count++
if count >= 100 {
wd.Close()
wd = intimate.GetChromeDriver(3030)
count = 0
log.Println("streamer find", len(articles))
if len(articles) == 0 {
sourceChannel.Operator = 5
sstore.UpdateOperator(sourceChannel)
}
counter.AddWithReset(1)
}
wd.Close()
wd.Quit()
}

View File

@@ -1,221 +0,0 @@
package main
import (
"database/sql"
"encoding/json"
"intimate"
"log"
"os"
"os/signal"
"regexp"
"sync/atomic"
"syscall"
"testing"
"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()
func TestCase(t *testing.T) {
var loop int32 = 1
wd := intimate.GetChromeDriver(3030)
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)
}()
var lasterr error = nil
// var err error
for atomic.LoadInt32(&loop) > 0 {
streamer, err := estore.Pop(intimate.Ptwitch, 0)
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 := updateUrl["live"]
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)
continue
}
streamer.LiveUrl = sql.NullString{String: liveUrl, Valid: true}
clog := &intimate.CollectLog{}
clog.UserId = streamer.UserId
time.Sleep(time.Millisecond * 500)
extractUserName(wd, streamer)
extractFollowers(wd, clog)
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 = string(streamer.Platform)
clog.UpdateTime = sql.NullTime{Time: time.Now(), Valid: true}
lastClogId := estore.InsertCollectLog(clog)
streamer.Operator = 100
streamer.LatestLogUid = lastClogId
estore.UpdateStreamer(streamer)
}
}
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 {
log.Println("label:", ltxt)
streamer.UserName = sql.NullString{String: ltxt, Valid: true}
return true, nil
}
}
return false, err
}, 6*time.Second)
}
func extractFollowers(wd selenium.WebDriver, clog *intimate.CollectLog) error {
return wd.WaitWithTimeout(func(web selenium.WebDriver) (bool, error) {
efollowers, err := web.FindElement(selenium.ByXPATH, "//div[@data-a-target='about-panel']//div[@class='tw-align-center']")
if err != nil {
return false, err
}
followers, err := efollowers.Text()
if err != nil || followers == "" {
return false, err
}
followers = regexp.MustCompile(`[\d,]+`).FindString(followers)
fint, _ := intimate.ParseNumber(followers)
clog.Followers = sql.NullInt64{Int64: int64(fint), Valid: true}
log.Println("followers: ", followers, fint)
return true, nil
}, 6*time.Second)
}
func extractViews(wd selenium.WebDriver, clog *intimate.CollectLog) error {
return wd.WaitWithTimeout(func(web selenium.WebDriver) (bool, error) {
views, err := web.FindElement(selenium.ByXPATH, "//a[@data-a-target='home-live-overlay-button']/span")
if views != nil {
if txt, err := views.Text(); err == nil {
vint, _ := intimate.ParseNumber(txt)
clog.Views = sql.NullInt64{Int64: vint, Valid: true}
log.Println("views:", txt)
views.Click()
extractTags(wd, clog)
extractTitle(wd, clog)
extractGratuity(wd, clog)
return true, nil
}
}
return false, err
}, time.Second*4)
}
func extractTitle(wd selenium.WebDriver, clog *intimate.CollectLog) error {
return wd.WaitWithTimeout(func(web selenium.WebDriver) (bool, 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}
return true, nil
}
}
return false, err
}, time.Second*4)
}
func extractTags(wd selenium.WebDriver, clog *intimate.CollectLog) error {
return wd.WaitWithTimeout(func(web selenium.WebDriver) (bool, error) {
tags, err := web.FindElements(selenium.ByXPATH, "//a[@aria-label and @data-a-target and @href]/div[@class and text()]")
if len(tags) == 0 {
return false, err
}
var stags []string
for _, tag := range tags {
if txt, err := tag.Text(); err == nil {
stags = append(stags, txt)
} else {
log.Println(err)
}
log.Println(tag.Text())
}
if len(stags) > 0 {
if tagbuf, err := json.Marshal(stags); err == nil {
clog.Tags = tagbuf
} else {
log.Println(err)
}
}
return true, nil
}, time.Second*4)
}
func extractGratuity(wd selenium.WebDriver, clog *intimate.CollectLog) error {
return wd.WaitWithTimeout(func(web selenium.WebDriver) (bool, error) {
btn, err := web.FindElement(selenium.ByXPATH, `//button[@data-test-selector="expand-grabber"]`)
if err == nil {
btn.Click()
time.Sleep(time.Second)
gifcount, err := web.FindElements(selenium.ByXPATH, `//div[@class="sub-gift-count tw-flex"]/p`)
if err == nil {
var gratuity int64 = 0
for _, gc := range gifcount {
if gtxt, err := gc.Text(); err == nil {
gint, _ := intimate.ParseNumber(gtxt)
gratuity += gint
} else {
log.Println(err)
}
}
clog.Gratuity = sql.NullInt64{Int64: gratuity, Valid: true}
}
return true, nil
}
return false, err
}, time.Second*4)
}