From 265c4e88c926102903418e90ad6f9d4c5670de6c Mon Sep 17 00:00:00 2001
From: filesite <release@filesite.io>
Date: Tue, 16 Apr 2024 18:05:50 +0800
Subject: [PATCH] add notify handler for HeroUnion task data save

---
 config.mjs          |  4 ++-
 lib/common.mjs      |  4 +++
 lib/taskMoniter.mjs | 72 +++++++++++++++++++++++++++++++++++++++++++++
 spider.mjs          | 12 ++++----
 4 files changed, 86 insertions(+), 6 deletions(-)

diff --git a/config.mjs b/config.mjs
index 3674833..ec09a02 100644
--- a/config.mjs
+++ b/config.mjs
@@ -14,7 +14,9 @@ export default {
         country: 'cn',                                      //爬虫所在国家
         lang: 'zh',                                         //爬虫支持的语言
         contact: 'https://filesite.io',                     //爬虫的联系方式
-        data_mode: 'json'                                   //爬虫支持的数据格式
+        data_mode: 'json',                                  //爬虫支持的数据格式
+
+        notify_max_try: 5                                   //爬虫完成任务回传数据最多尝试次数
     },
 
 
diff --git a/lib/common.mjs b/lib/common.mjs
index c7ff63b..4c4866e 100644
--- a/lib/common.mjs
+++ b/lib/common.mjs
@@ -9,6 +9,10 @@ export default {
         return today.toLocaleString(locales);
     },
 
+    getTimestampInSeconds: function() {
+        return Math.floor(Date.now() / 1000);
+    },
+
     removeFile: async function(filepath) {
         let done = false;
 
diff --git a/lib/taskMoniter.mjs b/lib/taskMoniter.mjs
index 7245191..c0b8566 100644
--- a/lib/taskMoniter.mjs
+++ b/lib/taskMoniter.mjs
@@ -20,7 +20,9 @@ import HeroBot from "./heroBot.mjs";
 class TaskMoniter {
     constructor(task_list_dir) {
         this.check_time_gap = 1;           //检测间隔时间，单位：分钟
+        this.notify_time_gap = 5;          //数据回调间隔时间，单位：分钟
         this.checking = false;
+        this.notifying = false;
 
         this.task_dir = task_list_dir;      //监控目录：任务列表保存目录
         this.tasks = {};                    //内存中的任务列表
@@ -150,6 +152,17 @@ class TaskMoniter {
         return true;
     }
 
+    updateTask(task_id, task) {
+        if (typeof(this.tasks[task.id]) == 'undefined') {
+            return false;
+        }
+
+        this.tasks[task.id] = task;
+
+        return true;
+    }
+
+    //检查新的数据抓取任务
     async checkTasks() {
         if (this.checking == true) {
             return;
@@ -191,6 +204,49 @@ class TaskMoniter {
         }
     }
 
+    //保存数据到HeroUnion联盟
+    //检查已经抓取到数据的任务
+    async notifyHandle(task) {
+        if (typeof(task.from) == 'undefined' || task.from != 'HeroUnion' || this.notifying) {
+            return false;
+        }
+
+        //已经完成回传
+        if (typeof(task.notified) != 'undefined' && task.notified) {
+            return false;
+        }
+
+        //判断当前任务数据回传次数是否小于最多尝试次数
+        if (typeof(task.notify_time) != 'undefined' && task.notify_time >= configs.herounion.notify_max_try) {
+            return false;
+        }
+
+        //尝试回传数据
+        this.notifying = true;
+        let saveRes = await this.heroBot.saveTaskData(task.id, task.token, task.data);
+        this.notifying = false;
+
+        if (typeof(task.notify_time) != 'undefined') {
+            task.notify_time ++;
+        }else {
+            task.notify_time = 1;                                //回传次数
+        }
+
+        task.notify_at = common.getTimestampInSeconds();     //回传时间戳
+
+        //如果返回数据code=1，则认为数据保存成功，否则过几分钟再次尝试
+        if (saveRes && saveRes.code == 1) {
+            task.notified = true;                            //记录已经完成回传
+            console.log("[%s][%s] Task %s's data save to HeroUnion done",
+                    common.getTimeString(), task.notify_time, task.id);
+        }else {
+            console.log("[%s][%s] Task %s's data save to HeroUnion failed, it will try again later.",
+                    common.getTimeString(), task.notify_time, task.id);
+        }
+
+        this.updateTask(task.id, task);                      //更新任务数据
+    }
+
     run() {     //开始监控任务目录，把所有任务缓存到内存
         console.log('[%s] TaskMoniter started.', common.getTimeString());
 
@@ -206,6 +262,22 @@ class TaskMoniter {
 
         task_auto_run.start();
         console.log('[%s] TaskMoniter auto check started.', common.getTimeString());
+
+
+        //定期向HeroUnion回传任务抓取结果
+        const task_notify_time = this.notify_time_gap;
+        const notify_auto_run = cron.schedule(`*/${task_notify_time} * * * *`, async () =>  {
+            let task = _self.tasks.find((item) => typeof(item.from) != 'undefined' && item.from == 'HeroUnion' && typeof(item.notified) == 'undefined');
+            if (task) {
+                console.log("[%s] Try to save task %s's data to HeroUnion", common.getTimeString(), task.id);
+                await _self.notifyHandle(task);
+            }
+        }, {
+            scheduled: false
+        });
+
+        notify_auto_run.start();
+        console.log('[%s] TaskMoniter auto notify started.', common.getTimeString());
     }
 }
 
diff --git a/spider.mjs b/spider.mjs
index e86b16e..e728214 100644
--- a/spider.mjs
+++ b/spider.mjs
@@ -75,16 +75,18 @@ import cron from 'node-cron';
             //console.log('Data got by bot', data);
 
             if (typeof(data.done) != 'undefined' && data.done == true) {
+                task.data = data;        //把抓取到的数据保存到任务里
+                taskMoniter.updateTask(task.id, task);
+
                 if (
                     await tajian.saveUrlShortcut(task.id, data)
                     && await tajian.saveDescriptionFiles(task.id, data)
                 ) {
-                    taskMoniter.setTaskDone(task.id);
+                    //马上回传一次数据
+                    taskMoniter.notifyHandle(task);
 
-                    //保存数据到HeroUnion联盟
-                    if (typeof(task.from) != 'undefined' && task.from == 'HeroUnion') {
-                        heroBot.saveTaskData(task.id, task.token, data);
-                    }
+                    //标记任务完成
+                    taskMoniter.setTaskDone(task.id);
                 }else {
                     taskMoniter.setTaskFailed(task.id);
                 }