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

@@ -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()
}