finish: 重构openrec

This commit is contained in:
eson 2020-08-17 13:10:29 +08:00
parent a50c879d83
commit d1298dc3f3
5 changed files with 84 additions and 164 deletions

View File

@ -15,6 +15,5 @@ import (
*/
func main() {
oe := &OpenrecExtractor{}
oe.Execute()
Execute()
}

View File

@ -3,11 +3,9 @@ package main
import (
"database/sql"
"encoding/json"
"errors"
"intimate"
"log"
"regexp"
"strconv"
"strings"
"time"
"github.com/474420502/extractor"
@ -17,20 +15,23 @@ import (
var estore = intimate.NewStoreExtractor()
var sstore = intimate.NewStoreSource(string(intimate.STOpenrec))
// OpenrecExtractor 提取方法
type OpenrecExtractor struct {
user *intimate.ExtractorSource
userLive *intimate.ExtractorSource
supporters *intimate.ExtractorSource
}
//UserInfo 提取信息的结构体
type UserInfo struct {
UserName string `exp:"//p[ contains(@class, 'c-global__user__profile__list__name__text')]"`
Followers int `exp:"//p[@class='c-global__user__count__row__right js-userCountFollowers']" mth:"r:ParseNumber"`
Views int `exp:"//ul[@class='c-contents']//p[@class='c-thumbnailVideo__footer__liveCount']" mth:"r:ExtractNumber"`
Followers int64 `exp:"//p[@class='c-global__user__count__row__right js-userCountFollowers']" mth:"r:ParseNumber"`
Views int64 `exp:"//ul[@class='c-contents']//p[@class='c-thumbnailVideo__footer__liveCount']" mth:"r:ExtractNumber"`
}
func (oe *OpenrecExtractor) Execute() {
//UserLive 提取信息的结构体
type UserLive struct {
Title string `exp:"//h1[contains(@class,'MovieTitle__Title')]"`
LiveStartTime string `exp:"//meta[@itemprop='uploadDate']/@content"`
LiveEndTime string `exp:"//meta[@itemprop='duration']/@content"`
Tags []string `exp:"//a[contains(@class,'TagButton')]"`
}
// Execute 执行
func Execute() {
ps := intimate.NewPerfectShutdown()
@ -47,6 +48,7 @@ func (oe *OpenrecExtractor) Execute() {
time.Sleep(time.Second * 5)
continue
}
lasterr = nil
sdata := source.Ext.([]byte)
datamap := gjson.ParseBytes(sdata).Map()
@ -56,31 +58,80 @@ func (oe *OpenrecExtractor) Execute() {
streamer := &intimate.Streamer{}
streamer.UserId = userId
// streamer.Platform = intimate.Popenrec
// streamer.Platform = intimate.Popenrec 不需要更新字段
htmlUser := datamap["html_user"]
oe.user = intimate.NewExtractorSource(&htmlUser)
oe.user.CreateExtractor()
userEtor := extractor.ExtractHtmlString(htmlUser.String())
log.Println(userEtor.GetObjectByTag(UserInfo{}))
ui, ok1 := userEtor.GetObjectByTag(UserInfo{}).(*UserInfo)
htmlLive := datamap["html_live"]
oe.userLive = intimate.NewExtractorSource(&htmlLive)
oe.userLive.CreateExtractor()
liveEtor := extractor.ExtractHtmlString(htmlLive.String())
ul, ok2 := liveEtor.GetObjectByTag(UserLive{}).(*UserLive)
jsonSupporters := datamap["json_supporters"]
oe.supporters = intimate.NewExtractorSource(&jsonSupporters)
clog := &intimate.CollectLog{}
// log.Println(anchorId)
if ok1 {
clog.Followers = sql.NullInt64{Int64: ui.Followers, Valid: true}
clog.Views = sql.NullInt64{Int64: ui.Views, Valid: true}
if ui.Views != 0 {
clog.IsLiveStreaming = true
}
streamer.UserName = sql.NullString{String: ui.UserName, Valid: true}
oe.extractFollowers(clog)
oe.extractUserName(streamer)
oe.extractViewsAndLiveStreaming(clog)
oe.extractGiversAndGratuity(clog)
oe.extractLive(clog)
oe.extractTags(clog)
giverjson := jsonSupporters
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.ErrorMsg = sql.NullString{String: err.Error(), Valid: true}
} else {
clog.Giver = giversbytes
}
clog.Gratuity = sql.NullInt64{Int64: gratuity, Valid: true}
} else {
log.Println("UserInfo may be not exists")
estore.UpdateError(streamer, errors.New("UserInfo may be not exists"))
continue
}
//log.Println(ul)
if ok2 {
clog.LiveTitle = sql.NullString{String: ul.Title, Valid: true}
startTime, err := time.ParseInLocation("2006-01-02T15:04:05Z07:00", ul.LiveStartTime, time.Local)
if err != nil {
log.Println(err)
} else {
clog.LiveStartTime = sql.NullTime{Time: startTime.Local(), Valid: true}
duration, err := intimate.ParseDuration(ul.LiveEndTime)
if err != nil {
log.Println(err)
} else {
endTime := startTime.Add(duration)
clog.LiveStartTime = sql.NullTime{Time: endTime.Local(), Valid: true}
}
}
if tags, err := json.Marshal(ul.Tags); err == nil {
clog.Tags = tags
} else {
log.Println("json error", ul.Tags, clog.Tags)
}
}
streamer.Uid = source.StreamerId.Int64
streamer.UpdateTime = source.UpdateTime
@ -89,6 +140,7 @@ func (oe *OpenrecExtractor) Execute() {
clog.Platform = intimate.Popenrec
clog.UserId = userId
clog.UpdateTime = source.UpdateTime
clog.StreamerUid = streamer.Uid
logUid := estore.InsertClog(clog)
@ -113,135 +165,3 @@ func (oe *OpenrecExtractor) Execute() {
}
}
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) extractUserName(streamer intimate.ISet) {
extractor := oe.user.GetExtractor()
xp, err := extractor.XPathResult("//p[ contains(@class, 'c-global__user__profile__list__name__text')]/text()")
if err != nil {
log.Println(err)
} else {
if xp.NodeIter().Next() {
userName := xp.String()
streamer.Set("UserName", sql.NullString{String: userName, Valid: true})
}
}
}
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", true)
}
}
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)
}
// log.Println(iter.Node().NodeValue(), tm.Local())
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])
}
tagsBytes, err := json.Marshal(tags)
if err != nil {
log.Println(err)
}
clog.Set("Tags", tagsBytes)
}

View File

@ -114,6 +114,5 @@ func TestUserName(t *testing.T) {
}
func TestExtractor(t *testing.T) {
oe := &OpenrecExtractor{}
oe.Execute()
Execute()
}

2
go.mod
View File

@ -3,7 +3,7 @@ module intimate
go 1.14
require (
github.com/474420502/extractor v0.9.4-0.20200814111732-bc270321f8f9
github.com/474420502/extractor v0.9.4-0.20200817020657-7d26da5d1e89
github.com/474420502/focus v0.12.0
github.com/474420502/gcurl v0.2.0
github.com/474420502/hunter v0.3.4

2
go.sum
View File

@ -8,6 +8,8 @@ github.com/474420502/extractor v0.9.3 h1:Cjri64DbgWQQ64EjPiBSQfUH9l0cYlzU8py0PQu
github.com/474420502/extractor v0.9.3/go.mod h1:Ss0KTfwsdB4XBpNda/V50rx21V9bl6/eQmyl50mjAS4=
github.com/474420502/extractor v0.9.4-0.20200814111732-bc270321f8f9 h1:cxgsTQwRJSiML4yBL40n/0pD/FbEqkCIXE7qq6hJyLg=
github.com/474420502/extractor v0.9.4-0.20200814111732-bc270321f8f9/go.mod h1:Ss0KTfwsdB4XBpNda/V50rx21V9bl6/eQmyl50mjAS4=
github.com/474420502/extractor v0.9.4-0.20200817020657-7d26da5d1e89 h1:6g4sPgooFdsVAdxNMhP6sqKQ0Z5EPBb4tGj9/absPoY=
github.com/474420502/extractor v0.9.4-0.20200817020657-7d26da5d1e89/go.mod h1:Ss0KTfwsdB4XBpNda/V50rx21V9bl6/eQmyl50mjAS4=
github.com/474420502/focus v0.12.0 h1:+icbmj7IEOefvTegHt5EpcHt6WFbe2miIrceUJx2Evo=
github.com/474420502/focus v0.12.0/go.mod h1:d0PMjtMxFz1a9HIhwyFPkWa+JF+0LgOrEUfd8iZka6s=
github.com/474420502/gcurl v0.1.2 h1:ON9Yz3IgAdtDlFlHfkAJ3aIEBDxH0RiViPE5ST5ohKg=