From 41e973a4d3d4686b0e58535335f0642649811d71 Mon Sep 17 00:00:00 2001
From: Marvin Zhang <tikazyq@163.com>
Date: Sat, 21 Dec 2024 15:33:07 +0800
Subject: [PATCH] fix: misaligned nodes when running tasks from a schedule
 through enhancement by adding new route and handler for running schedules

- Introduced a new POST endpoint '/:id/run' in the router to trigger the execution of a schedule.
- Implemented the PostScheduleRun function to handle the execution logic, including validation of input parameters and scheduling tasks.
- Refactored the PostSpiderRun function to streamline the scheduling process by directly calling the admin service.
- Updated the FsFileInfo struct to clarify the children field description.
- Modified the Task model to make NodeIds field optional in JSON serialization.
---
 core/controllers/router.go   |  5 ++++
 core/controllers/schedule.go | 48 ++++++++++++++++++++++++++++++++++++
 core/controllers/spider.go   |  6 ++---
 core/controllers/task.go     |  6 +++++
 core/entity/fs_file_info.go  |  2 +-
 core/models/models/task.go   |  2 +-
 6 files changed, 63 insertions(+), 6 deletions(-)

diff --git a/core/controllers/router.go b/core/controllers/router.go
index f02299f5..ce4bf778 100644
--- a/core/controllers/router.go
+++ b/core/controllers/router.go
@@ -101,6 +101,11 @@ func InitRoutes(app *gin.Engine) (err error) {
 			Path:        "/:id/disable",
 			HandlerFunc: PostScheduleDisable,
 		},
+		{
+			Method:      http.MethodPost,
+			Path:        "/:id/run",
+			HandlerFunc: PostScheduleRun,
+		},
 	}...))
 	RegisterController(groups.AuthGroup, "/spiders", NewController[models.Spider]([]Action{
 		{
diff --git a/core/controllers/schedule.go b/core/controllers/schedule.go
index 74b37aac..a3a0056d 100644
--- a/core/controllers/schedule.go
+++ b/core/controllers/schedule.go
@@ -2,9 +2,11 @@ package controllers
 
 import (
 	errors2 "errors"
+	"github.com/crawlab-team/crawlab/core/interfaces"
 	"github.com/crawlab-team/crawlab/core/models/models"
 	"github.com/crawlab-team/crawlab/core/models/service"
 	"github.com/crawlab-team/crawlab/core/schedule"
+	"github.com/crawlab-team/crawlab/core/spider/admin"
 	"github.com/gin-gonic/gin"
 	"go.mongodb.org/mongo-driver/bson/primitive"
 )
@@ -116,3 +118,49 @@ func postScheduleEnableDisableFunc(isEnable bool) func(c *gin.Context) {
 		HandleSuccess(c)
 	}
 }
+
+func PostScheduleRun(c *gin.Context) {
+	id, err := primitive.ObjectIDFromHex(c.Param("id"))
+	if err != nil {
+		HandleErrorBadRequest(c, err)
+		return
+	}
+
+	// options
+	var opts interfaces.SpiderRunOptions
+	if err := c.ShouldBindJSON(&opts); err != nil {
+		HandleErrorInternalServerError(c, err)
+		return
+	}
+	if opts.ScheduleId.IsZero() {
+		opts.ScheduleId = id
+	}
+
+	// schedule
+	sch, err := service.NewModelService[models.Schedule]().GetById(id)
+	if err != nil {
+		HandleErrorInternalServerError(c, err)
+		return
+	}
+
+	// spider
+	s, err := service.NewModelService[models.Spider]().GetById(sch.SpiderId)
+	if err != nil {
+		HandleErrorInternalServerError(c, err)
+		return
+	}
+
+	// user
+	if u := GetUserFromContext(c); u != nil {
+		opts.UserId = u.GetId()
+	}
+
+	// schedule tasks
+	taskIds, err := admin.GetSpiderAdminService().Schedule(s.Id, &opts)
+	if err != nil {
+		HandleErrorInternalServerError(c, err)
+		return
+	}
+
+	HandleSuccessWithData(c, taskIds)
+}
diff --git a/core/controllers/spider.go b/core/controllers/spider.go
index 163e6c53..36746166 100644
--- a/core/controllers/spider.go
+++ b/core/controllers/spider.go
@@ -660,10 +660,8 @@ func PostSpiderRun(c *gin.Context) {
 		opts.UserId = u.GetId()
 	}
 
-	adminSvc := admin.GetSpiderAdminService()
-
-	// schedule
-	taskIds, err := adminSvc.Schedule(id, &opts)
+	// schedule tasks
+	taskIds, err := admin.GetSpiderAdminService().Schedule(id, &opts)
 	if err != nil {
 		HandleErrorInternalServerError(c, err)
 		return
diff --git a/core/controllers/task.go b/core/controllers/task.go
index 1d2964f6..7032a476 100644
--- a/core/controllers/task.go
+++ b/core/controllers/task.go
@@ -339,6 +339,12 @@ func PostTaskRestart(c *gin.Context) {
 		Priority: t.Priority,
 	}
 
+	// normalize options for tasks with assigned node
+	if !t.NodeId.IsZero() {
+		opts.NodeIds = []primitive.ObjectID{t.NodeId}
+		opts.Mode = constants.RunTypeSelectedNodes
+	}
+
 	// user
 	if u := GetUserFromContext(c); u != nil {
 		opts.UserId = u.Id
diff --git a/core/entity/fs_file_info.go b/core/entity/fs_file_info.go
index 826ec131..c2adde5f 100644
--- a/core/entity/fs_file_info.go
+++ b/core/entity/fs_file_info.go
@@ -13,7 +13,7 @@ type FsFileInfo struct {
 	Extension string                  `json:"extension"` // file extension
 	IsDir     bool                    `json:"is_dir"`    // whether it is directory
 	FileSize  int64                   `json:"file_size"` // file size (bytes)
-	Children  []interfaces.FsFileInfo `json:"children"`  // children for sub-directory
+	Children  []interfaces.FsFileInfo `json:"children"`  // children for subdirectory
 	ModTime   time.Time               `json:"mod_time"`  // modification time
 	Mode      os.FileMode             `json:"mode"`      // file mode
 	Hash      string                  `json:"hash"`      // file hash
diff --git a/core/models/models/task.go b/core/models/models/task.go
index 9fa2083e..69b9c78b 100644
--- a/core/models/models/task.go
+++ b/core/models/models/task.go
@@ -17,8 +17,8 @@ type Task struct {
 	ScheduleId      primitive.ObjectID   `json:"schedule_id" bson:"schedule_id"`
 	Type            string               `json:"type" bson:"type"`
 	Mode            string               `json:"mode" bson:"mode"`
-	NodeIds         []primitive.ObjectID `json:"node_ids" bson:"node_ids"`
 	Priority        int                  `json:"priority" bson:"priority"`
+	NodeIds         []primitive.ObjectID `json:"node_ids,omitempty" bson:"-"`
 	Stat            *TaskStat            `json:"stat,omitempty" bson:"-"`
 	Spider          *Spider              `json:"spider,omitempty" bson:"-"`
 	Schedule        *Schedule            `json:"schedule,omitempty" bson:"-"`