git-sync/crawlab
1
0
Fork 0
mirror of https://github.com/crawlab-team/crawlab.git synced 2026-01-23 17:31:11 +01:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #6 from wo10378931/develop

Browse Source 
Develop
This commit is contained in:
暗音
2019-10-16 16:06:50 +08:00
committed by GitHub
parent 344ae0ed1b c028c4d1f3
commit c2d69d7075
10 changed files with 275 additions and 210 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
backend/model/node.go
View File

@@ -4,6 +4,7 @@ import (
"crawlab/constants"
"crawlab/database"
"crawlab/services/register"
"errors"
"github.com/apex/log"
"github.com/globalsign/mgo"
"github.com/globalsign/mgo/bson"
@@ -158,16 +159,19 @@ func GetNodeList(filter interface{}) ([]Node, error) {
func GetNode(id bson.ObjectId) (Node, error) {
var node Node
if id.Hex() == "" {
return node, nil
log.Infof("id is empty")
debug.PrintStack()
return node, errors.New("id is empty")
}
s, c := database.GetCol("nodes")
defer s.Close()
if err := c.FindId(id).One(&node); err != nil {
if err != mgo.ErrNotFound {
log.Errorf(err.Error())
debug.PrintStack()
}
log.Errorf(err.Error())
debug.PrintStack()
return node, err
}
return node, nil

31
backend/model/schedule.go
View File

@@ -45,6 +45,27 @@ func (sch *Schedule) Delete() error {
return c.RemoveId(sch.Id)
}
func (sch *Schedule) SyncNodeIdAndSpiderId(node Node, spider Spider) {
sch.syncNodeId(node)
sch.syncSpiderId(spider)
}
func (sch *Schedule) syncNodeId(node Node) {
if node.Id.Hex() == sch.NodeId.Hex() {
return
}
sch.NodeId = node.Id
_ = sch.Save()
}
func (sch *Schedule) syncSpiderId(spider Spider) {
if spider.Id.Hex() == sch.SpiderId.Hex() {
return
}
sch.SpiderId = spider.Id
_ = sch.Save()
}
func GetScheduleList(filter interface{}) ([]Schedule, error) {
s, c := database.GetCol("schedules")
defer s.Close()
@@ -103,13 +124,11 @@ func UpdateSchedule(id bson.ObjectId, item Schedule) error {
if err := c.FindId(id).One(&result); err != nil {
return err
}
node, err := GetNode(item.NodeId)
if err != nil {
log.Errorf("get node error: %s", err.Error())
debug.PrintStack()
return nil
return err
}
item.NodeKey = node.Key
if err := item.Save(); err != nil {
return err
@@ -123,9 +142,7 @@ func AddSchedule(item Schedule) error {
node, err := GetNode(item.NodeId)
if err != nil {
log.Errorf("get node error: %s", err.Error())
debug.PrintStack()
return nil
return err
}
item.Id = bson.NewObjectId()

22
backend/routes/schedule.go
View File

@@ -45,13 +45,14 @@ func PostSchedule(c *gin.Context) {
HandleError(http.StatusBadRequest, c, err)
return
}
// 验证cron表达式
if err := services.ParserCron(newItem.Cron); err != nil {
HandleError(http.StatusOK, c, err)
return
}
newItem.Id = bson.ObjectIdHex(id)
// 如果node_id为空，则置为空ObjectId
//if newItem.NodeId == "" {
// newItem.NodeId = bson.ObjectIdHex(constants.ObjectIdNull)
//}
// 更新数据库
if err := model.UpdateSchedule(bson.ObjectIdHex(id), newItem); err != nil {
HandleError(http.StatusInternalServerError, c, err)
@@ -79,10 +80,11 @@ func PutSchedule(c *gin.Context) {
return
}
// 如果node_id为空，则置为空ObjectId
//if item.NodeId == "" {
// item.NodeId = bson.ObjectIdHex(constants.ObjectIdNull)
//}
// 验证cron表达式
if err := services.ParserCron(item.Cron); err != nil {
HandleError(http.StatusOK, c, err)
return
}
// 更新数据库
if err := model.AddSchedule(item); err != nil {

16
backend/services/node.go
View File

@@ -100,13 +100,13 @@ func handleNodeInfo(key string, data Data) {
defer s.Close()
// 同个key可能因为并发，被注册多次
//var nodes []model.Node
//_ = c.Find(bson.M{"key": key}).All(&nodes)
//if nodes != nil && len(nodes) > 1 {
// for _, node := range nodes {
// _ = c.RemoveId(node.Id)
// }
//}
var nodes []model.Node
_ = c.Find(bson.M{"key": key}).All(&nodes)
if nodes != nil && len(nodes) > 1 {
for _, node := range nodes {
_ = c.RemoveId(node.Id)
}
}
var node model.Node
if err := c.Find(bson.M{"key": key}).One(&node); err != nil {
@@ -227,7 +227,7 @@ func InitNodeService() error {
}
// 首次更新节点数据（注册到Redis）
// UpdateNodeData()
UpdateNodeData()
// 获取当前节点
node, err := model.GetCurrentNode()

17
backend/services/schedule.go
View File

@@ -5,7 +5,7 @@ import (
"crawlab/lib/cron"
"crawlab/model"
"github.com/apex/log"
uuid "github.com/satori/go.uuid"
"github.com/satori/go.uuid"
"runtime/debug"
)
@@ -31,6 +31,9 @@ func AddTask(s model.Schedule) func() {
return
}
// 同步ID到定时任务
s.SyncNodeIdAndSpiderId(node, *spider)
// 生成任务ID
id := uuid.NewV4()
@@ -119,6 +122,18 @@ func (s *Scheduler) RemoveAll() {
}
}
// 验证cron表达式是否正确
func ParserCron(spec string) error {
parser := cron.NewParser(
cron.Second | cron.Minute | cron.Hour | cron.Dom | cron.Month | cron.Dow | cron.Descriptor,
)
if _, err := parser.Parse(spec); err != nil {
return err
}
return nil
}
func (s *Scheduler) Update() error {
// 删除所有定时任务
s.RemoveAll()

231
backend/services/task.go
View File

@@ -100,91 +100,85 @@ func AssignTask(task model.Task) error {
return nil
}
// 执行shell命令
func ExecuteShellCmd(cmdStr string, cwd string, t model.Task, s model.Spider) (err error) {
log.Infof("cwd: " + cwd)
log.Infof("cmd: " + cmdStr)
// 设置环境变量
func SetEnv(cmd *exec.Cmd, envs []model.Env, taskId string, dataCol string) *exec.Cmd {
// 默认环境变量
cmd.Env = append(os.Environ(), "CRAWLAB_TASK_ID="+taskId)
cmd.Env = append(cmd.Env, "CRAWLAB_COLLECTION="+dataCol)
cmd.Env = append(cmd.Env, "PYTHONUNBUFFERED=0")
cmd.Env = append(cmd.Env, "PYTHONIOENCODING=utf-8")
cmd.Env = append(cmd.Env, "TZ=Asia/Shanghai")
// 生成执行命令
var cmd *exec.Cmd
if runtime.GOOS == constants.Windows {
cmd = exec.Command("cmd", "/C", cmdStr)
} else {
cmd = exec.Command("sh", "-c", cmdStr)
}
// 工作目录
cmd.Dir = cwd
// 指定stdout, stderr日志位置
fLog, err := os.Create(t.LogPath)
if err != nil {
HandleTaskError(t, err)
return err
}
defer fLog.Close()
cmd.Stdout = fLog
cmd.Stderr = fLog
// 添加默认环境变量
cmd.Env = append(cmd.Env, "CRAWLAB_TASK_ID="+t.Id)
cmd.Env = append(cmd.Env, "CRAWLAB_COLLECTION="+s.Col)
// 添加任务环境变量
for _, env := range s.Envs {
//任务环境变量
for _, env := range envs {
cmd.Env = append(cmd.Env, env.Name+"="+env.Value)
}
// 起一个goroutine来监控进程
ch := utils.TaskExecChanMap.ChanBlocked(t.Id)
go func() {
// 传入信号，此处阻塞
signal := <-ch
log.Infof("cancel process signal: %s", signal)
if signal == constants.TaskCancel && cmd.Process != nil {
// 取消进程
if err := syscall.Kill(-cmd.Process.Pid, syscall.SIGKILL); err != nil {
log.Errorf("process kill error: %s", err.Error())
debug.PrintStack()
}
t.Status = constants.StatusCancelled
t.Error = "user kill the process ..."
} else {
// 保存任务
t.Status = constants.StatusFinished
}
t.FinishTs = time.Now()
if err := t.Save(); err != nil {
log.Infof("save task error: %s", err.Error())
// TODO 全局环境变量
return cmd
}
func SetLogConfig(cmd *exec.Cmd, path string) error {
fLog, err := os.Create(path)
if err != nil {
log.Errorf("create task log file error: %s", path)
debug.PrintStack()
return err
}
cmd.Stdout = fLog
cmd.Stderr = fLog
return nil
}
func FinishOrCancelTask(ch chan string, cmd *exec.Cmd, t model.Task) {
// 传入信号，此处阻塞
signal := <-ch
log.Infof("process received signal: %s", signal)
if signal == constants.TaskCancel && cmd.Process != nil {
// 取消进程
if err := syscall.Kill(-cmd.Process.Pid, syscall.SIGKILL); err != nil {
log.Errorf("process kill error: %s", err.Error())
debug.PrintStack()
return
t.Error = "kill process error: " + err.Error()
t.Status = constants.StatusError
} else {
t.Error = "user kill the process ..."
t.Status = constants.StatusCancelled
}
}()
} else {
// 保存任务
t.Status = constants.StatusFinished
}
// 在选择所有节点执行的时候，实际就是随机一个节点执行的，
cmd.SysProcAttr = &syscall.SysProcAttr{Setpgid: true}
t.FinishTs = time.Now()
_ = t.Save()
}
// 异步启动进程
func StartTaskProcess(cmd *exec.Cmd, t model.Task) error {
if err := cmd.Start(); err != nil {
log.Errorf("start spider error:{}", err.Error())
debug.PrintStack()
return err
}
// 保存pid到task
t.Pid = cmd.Process.Pid
if err := t.Save(); err != nil {
log.Errorf("save task pid error: %s", err.Error())
debug.PrintStack()
t.Error = "start task error: " + err.Error()
t.Status = constants.StatusError
t.FinishTs = time.Now()
_ = t.Save()
return err
}
// 同步等待进程完成
return nil
}
func WaitTaskProcess(cmd *exec.Cmd, t model.Task) error {
if err := cmd.Wait(); err != nil {
log.Errorf("wait process finish error: %s", err.Error())
debug.PrintStack()
if exitError, ok := err.(*exec.ExitError); ok {
exitCode := exitError.ExitCode()
log.Errorf("exit error, exit code: %d", exitCode)
// 非kill 的错误类型
if exitCode != -1 {
// 非手动kill保存为错误状态
@@ -194,6 +188,52 @@ func ExecuteShellCmd(cmdStr string, cwd string, t model.Task, s model.Spider) (e
_ = t.Save()
}
}
return err
}
return nil
}
// 执行shell命令
func ExecuteShellCmd(cmdStr string, cwd string, t model.Task, s model.Spider) (err error) {
log.Infof("cwd: %s", cwd)
log.Infof("cmd: %s", cmdStr)
// 生成执行命令
var cmd *exec.Cmd
if runtime.GOOS == constants.Windows {
cmd = exec.Command("cmd", "/C", cmdStr)
} else {
cmd = exec.Command("")
cmd = exec.Command("sh", "-c", cmdStr)
}
// 工作目录
cmd.Dir = cwd
// 日志配置
if err := SetLogConfig(cmd, t.LogPath); err != nil {
return err
}
// 环境变量配置
cmd = SetEnv(cmd, s.Envs, t.Id, s.Col)
// 起一个goroutine来监控进程
ch := utils.TaskExecChanMap.ChanBlocked(t.Id)
go FinishOrCancelTask(ch, cmd, t)
// kill的时候，可以kill所有的子进程
cmd.SysProcAttr = &syscall.SysProcAttr{Setpgid: true}
// 启动进程
if err := StartTaskProcess(cmd, t); err != nil {
return err
}
// 同步等待进程完成
if err := WaitTaskProcess(cmd, t); err != nil {
return err
}
ch <- constants.TaskFinish
@@ -208,6 +248,7 @@ func MakeLogDir(t model.Task) (fileDir string, err error) {
// 如果日志目录不存在，生成该目录
if !utils.Exists(fileDir) {
if err := os.MkdirAll(fileDir, 0777); err != nil {
log.Errorf("execute task, make log dir error: %s", err.Error())
debug.PrintStack()
return "", err
}
@@ -272,85 +313,55 @@ func ExecuteTask(id int) {
// 获取当前节点
node, err := model.GetCurrentNode()
if err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
log.Errorf("execute task get current node error: %s", err.Error())
debug.PrintStack()
return
}
// 公共队列
queuePub := "tasks:public"
// 节点队列
queueCur := "tasks:node:" + node.Id.Hex()
// 节点队列任务
var msg string
msg, err = database.RedisClient.LPop(queueCur)
if msg != "" {
log.Infof("queue cur: %s", msg)
}
if err != nil {
if msg == "" {
// 节点队列没有任务，获取公共队列任务
msg, err = database.RedisClient.LPop(queuePub)
if err != nil {
if msg == "" {
// 公共队列没有任务
log.Debugf(GetWorkerPrefix(id) + "没有任务...")
return
} else {
log.Errorf(GetWorkerPrefix(id) + err.Error())
debug.PrintStack()
return
}
}
} else {
log.Errorf(GetWorkerPrefix(id) + err.Error())
debug.PrintStack()
return
if msg, err = database.RedisClient.LPop(queueCur); err != nil {
// 节点队列没有任务，获取公共队列任务
queuePub := "tasks:public"
if msg, err = database.RedisClient.LPop(queuePub); err != nil {
}
}
if msg == "" {
return
}
// 反序列化
tMsg := TaskMessage{}
if err := json.Unmarshal([]byte(msg), &tMsg); err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
debug.PrintStack()
log.Errorf("json string to struct error: %s", err.Error())
return
}
// 获取任务
t, err := model.GetTask(tMsg.Id)
if err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
log.Errorf("execute task, get task error: %s", err.Error())
return
}
// 获取爬虫
spider, err := t.GetSpider()
if err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
log.Errorf("execute task, get spider error: %s", err.Error())
return
}
// 创建日志目录
fileDir, err := MakeLogDir(t)
if err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
var fileDir string
if fileDir, err = MakeLogDir(t); err != nil {
return
}
// 获取日志文件路径
t.LogPath = GetLogFilePaths(fileDir)
// 创建日志目录文件夹
fileStdoutDir := filepath.Dir(t.LogPath)
if !utils.Exists(fileStdoutDir) {
if err := os.MkdirAll(fileStdoutDir, os.ModePerm); err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
return
}
}
// 工作目录
cwd := filepath.Join(
viper.GetString("spider.path"),

21
frontend/src/components/Common/CrawlConfirmDialog.vue
View File

@@ -9,9 +9,8 @@
<el-form label-width="80px">
<el-form-item :label="$t('Node')">
<el-select v-model="nodeId">
<el-option value="" :label="$t('All Nodes')"/>
<el-option
v-for="op in $store.state.node.nodeList"
v-for="op in nodeList"
:key="op._id"
:value="op._id"
:disabled="op.status !== 'online'"
@@ -31,6 +30,7 @@
</template>
<script>
import request from '../../api/request'
export default {
name: 'CrawlConfirmDialog',
props: {
@@ -46,7 +46,8 @@ export default {
data () {
return {
nodeId: '',
param: ''
param: '',
nodeList: []
}
},
methods: {
@@ -61,6 +62,20 @@ export default {
this.$emit('close')
this.$st.sendEv('爬虫', '运行')
}
},
created () {
// 节点列表
request.get('/nodes', {}).then(response => {
this.nodeList = response.data.data.map(d => {
d.systemInfo = {
os: '',
arch: '',
num_cpu: '',
executables: []
}
return d
})
})
}
}
</script>

101
frontend/src/views/schedule/ScheduleList.vue
View File

@@ -14,7 +14,7 @@
<el-form-item :label="$t('Schedule Name')" prop="name" required>
<el-input v-model="scheduleForm.name" :placeholder="$t('Schedule Name')"></el-input>
</el-form-item>
<el-form-item :label="$t('Node')" prop="node_id">
<el-form-item :label="$t('Node')" prop="node_id" required>
<el-select v-model="scheduleForm.node_id">
<!--<el-option :label="$t('All Nodes')" value="000000000000000000000000"></el-option>-->
<el-option
@@ -38,21 +38,22 @@
</el-option>
</el-select>
</el-form-item>
<el-form-item :label="$t('schedules.cron')" prop="cron" :rules="cronRules" required>
<template slot="label">
<el-tooltip :content="$t('schedules.cron_format')"
placement="top">
<span>
{{$t('schedules.cron')}}
<i class="fa fa-exclamation-circle"></i>
</span>
</el-tooltip>
</template>
<el-input style="width:calc(100% - 100px);padding-right:10px"
<!--:rules="cronRules"-->
<el-form-item :label="$t('schedules.cron')" prop="cron" required>
<!--<template slot="label">-->
<!--<el-tooltip :content="$t('schedules.cron_format')"-->
<!--placement="top">-->
<!--<span>-->
<!--{{$t('schedules.cron')}}-->
<!--<i class="fa fa-exclamation-circle"></i>-->
<!--</span>-->
<!--</el-tooltip>-->
<!--</template>-->
<el-input style="padding-right:10px"
v-model="scheduleForm.cron"
:placeholder="$t('schedules.cron')">
</el-input>
<el-button size="small" style="width:100px" type="primary" @click="onShowCronDialog">{{$t('schedules.add_cron')}}</el-button>
<!--<el-button size="small" style="width:100px" type="primary" @click="onShowCronDialog">{{$t('schedules.add_cron')}}</el-button>-->
</el-form-item>
<el-form-item :label="$t('Execute Command')" prop="params">
<el-input v-model="spider.cmd"
@@ -69,6 +70,7 @@
:placeholder="$t('Schedule Description')"></el-input>
</el-form-item>
</el-form>
<!--取消、保存-->
<span slot="footer" class="dialog-footer">
<el-button size="small" @click="onCancel">{{$t('Cancel')}}</el-button>
<el-button size="small" type="primary" @click="onAddSubmit">{{$t('Submit')}}</el-button>
@@ -76,9 +78,9 @@
</el-dialog>
<!--cron generation popup-->
<el-dialog title="生成 Cron" :visible.sync="showCron">
<vcrontab @hide="showCron=false" @fill="onCrontabFill" :expression="expression"></vcrontab>
</el-dialog>
<!--<el-dialog title="生成 Cron" :visible.sync="showCron">-->
<!--<vcrontab @hide="showCron=false" @fill="onCrontabFill" :expression="expression"></vcrontab>-->
<!--</el-dialog>-->
<el-card style="border-radius: 0">
<!--filter-->
@@ -131,7 +133,7 @@
</template>
<script>
import vcrontab from 'vcrontab'
// import vcrontab from 'vcrontab'
import request from '../../api/request'
import {
mapState
@@ -139,21 +141,7 @@ import {
export default {
name: 'ScheduleList',
components: { vcrontab },
data () {
const cronValidator = (rule, value, callback) => {
let patArr = []
for (let i = 0; i < 6; i++) {
patArr.push('[/*,0-9-]+')
}
const pat = '^' + patArr.join(' ') + '( [/*,0-9-]+)?' + '$'
if (!value) {
callback(new Error('cron cannot be empty'))
} else if (!value.match(pat)) {
callback(new Error('cron format is invalid'))
}
callback()
}
return {
columns: [
{ name: 'name', label: 'Name', width: '180' },
@@ -166,12 +154,10 @@ export default {
isEdit: false,
dialogTitle: '',
dialogVisible: false,
cronRules: [
{ validator: cronValidator, trigger: 'blur' }
],
showCron: false,
expression: '',
spiderList: []
spiderList: [],
nodeList: []
}
},
computed: {
@@ -179,9 +165,6 @@ export default {
'scheduleList',
'scheduleForm'
]),
...mapState('node', [
'nodeList'
]),
filteredTableData () {
return this.scheduleList
},
@@ -210,19 +193,25 @@ export default {
onAddSubmit () {
this.$refs.scheduleForm.validate(res => {
if (res) {
let action
if (this.isEdit) {
action = 'editSchedule'
} else {
action = 'addSchedule'
}
this.$store.dispatch('schedule/' + action, this.scheduleForm._id)
.then(() => {
request.post(`/schedules/${this.scheduleForm._id}`, this.scheduleForm).then(response => {
if (response.data.error) {
this.$message.error(response.data.error)
return
}
this.dialogVisible = false
setTimeout(() => {
this.$store.dispatch('schedule/getScheduleList')
}, 100)
this.$store.dispatch('schedule/getScheduleList')
})
} else {
request.put('/schedules', this.scheduleForm).then(response => {
if (response.data.error) {
this.$message.error(response.data.error)
return
}
this.dialogVisible = false
this.$store.dispatch('schedule/getScheduleList')
})
}
}
})
this.$st.sendEv('定时任务', '提交')
@@ -268,7 +257,21 @@ export default {
},
created () {
this.$store.dispatch('schedule/getScheduleList')
this.$store.dispatch('node/getNodeList')
// 节点列表
request.get('/nodes', {}).then(response => {
this.nodeList = response.data.data.map(d => {
d.systemInfo = {
os: '',
arch: '',
num_cpu: '',
executables: []
}
return d
})
})
// 爬虫列表
request.get('/spiders', {})
.then(response => {
this.spiderList = response.data.data.list

7
frontend/src/views/spider/SpiderList.vue
View File

@@ -429,13 +429,6 @@ export default {
} else {
return false
}
// if (this.isCustomized(row)) {
// // customized spider
// return !!row.cmd
// } else {
// // configurable spider
// return !!row.fields
// }
},
isCustomized (row) {
return row.type === 'customized'

25
frontend/src/views/task/TaskDetail.vue
View File

@@ -35,6 +35,7 @@ import {
import TaskOverview from '../../components/Overview/TaskOverview'
import GeneralTableView from '../../components/TableView/GeneralTableView'
import LogView from '../../components/ScrollView/LogView'
import request from '../../api/request'
export default {
name: 'TaskDetail',
@@ -46,12 +47,12 @@ export default {
data () {
return {
activeTabName: 'overview',
handle: undefined
handle: undefined,
taskLog: ''
}
},
computed: {
...mapState('task', [
'taskLog',
'taskResultsData',
'taskResultsTotalCount'
]),
@@ -97,18 +98,22 @@ export default {
downloadCSV () {
this.$store.dispatch('task/getTaskResultExcel', this.$route.params.id)
this.$st.sendEv('任务详情-结果', '下载CSV')
},
getTaskLog () {
if (this.$route.params.id) {
request.get(`/tasks/${this.$route.params.id}/log`).then(response => {
this.taskLog = response.data.data
})
}
}
},
async created () {
await this.$store.dispatch('task/getTaskData', this.$route.params.id)
this.$store.dispatch('task/getTaskLog', this.$route.params.id)
created () {
this.$store.dispatch('task/getTaskData', this.$route.params.id)
this.$store.dispatch('task/getTaskResults', this.$route.params.id)
if (this.taskForm && ['running'].includes(this.taskForm.status)) {
this.handle = setInterval(() => {
this.$store.dispatch('task/getTaskLog', this.$route.params.id)
}, 5000)
}
this.handle = setInterval(() => {
this.getTaskLog()
}, 5000)
},
destroyed () {
clearInterval(this.handle)