feat(contentjson): add judge the json of content

build.sh

@@ -2,4 +2,5 @@ rm build/ -rf
 mkdir -p ./build 
 cd proxyserver && go build && mv proxyserver ../build/ && cd ..
 cp start.sh build/start.sh
+cp stop.sh build/stop.sh
 cp proxyserver/config.yaml build/config.yaml

chromeproxy/.vscode/launch.json

@@ -14,6 +14,7 @@
                 "--ignore-certificate-errors",
                 "--disable-dev-shm-usage",
                 "--mute-audio",
+                "--single-process",
                 "--safebrowsing-disable-auto-update", 
                 "--disable-gpu",
                 "--no-sandbox",

chromeproxy/background/capture.js

@@ -1,23 +1,4 @@
-
-var href = window.location.href;
-var content = document.documentElement.innerHTML;
-
-setTimeout(function(){
-    window.close();
-}, 15000)
-
-if (condition == undefined) {
-    condition = function () {
-        return true;
-    };
-}
-
-if (condition()) {
-    Tell(BackgroundMsgType.CONTENT, content);
-} else {
-    Tell(BackgroundMsgType.NOTWANT, content);
-}
-
+ 
 function Tell(backgroundType, content) {
     try {
         if (content == undefined) {
@@ -29,4 +10,31 @@ function Tell(backgroundType, content) {
         console.log(error);
         window.location.href = Host;
     }
-}
+}
+ 
+try {
+    if (condition == undefined) {
+        condition = function () {
+            return true;
+        };
+    }
+} catch (error) {
+    condition = function () {
+        return true;
+    };
+}
+
+var href = window.location.href;
+var content = "";
+if(document.contentType == "application/json") {
+    content = document.documentElement.innerText;
+} else {
+    content = document.documentElement.innerHTML;
+}
+
+if (condition()) {
+    Tell(BackgroundMsgType.CONTENT, content);
+} else {
+    Tell(BackgroundMsgType.NOTWANT, content);
+}
+

chromeproxy/background/worker.js

@@ -16,8 +16,29 @@ chrome.runtime.onMessage.addListener(function (request, sender, sendResponse) {
       GetTask(sender);
       break;
     case BackgroundMsgType.NOTWANT:
-      ErrorTask(sender, request.content);
+      task = task_manager[sender.tab.id];
+      
+      if(task.data.waitcapture != undefined) {
+        setTimeout(function(){
+          CaptureContentCurrentPage(sender.tab.id);
+        }, task.data.waitcapture);
+      } else {
+
+        task.data.retry --;
+        if(task.data.retry >= 0) {
+          setTimeout(function(){
+            CaptureContent(task);
+          }, 6000)
+        } else {
+          ErrorTask(sender, request.content);
+        }
+      }
+
+ 
       break;
+    case BackgroundMsgType.CLOSETAB:
+      delete task_manager[sender.tab.id] 
+      chrome.tabs.remove(sender.tab.id);
     default:
       break;
   }
@@ -39,19 +60,33 @@ function GetTask(sender) {
   })
 }
 
+function CaptureContentCurrentPage(tabid) {
+  chrome.tabs.executeScript(tabid, { runAt: "document_end", file: "base.js" }, function(){
+    chrome.tabs.executeScript(tabid, { runAt: "document_end", file: "background/capture.js" });
+  });
+}
+
 function CaptureContent(task) {
   if (task.code == 200) {
+
+    if(task.data.retry == undefined) {
+      task.data.retry = 1;
+    }
+
     chrome.tabs.create({ url: task.data.url }, function (tab) {
       task_manager[tab.id] = task;
       if (task.data.content_condition) {
-        condition = `${task.data.content_condition}`
+        condition = `${task.data.content_condition}`;
+      } else {
+        condition = undefined;
       }
-      if (condition) {
+
+      if (condition != undefined) {
         chrome.tabs.executeScript(tab.id, { runAt: "document_end", code: condition }, function () {
-          chrome.tabs.executeScript(tab.id, { runAt: "document_end", file: "background/capture.js" });
+          CaptureContentCurrentPage(tab.id)
         });
       } else {
-        chrome.tabs.executeScript(tab.id, { runAt: "document_end", file: "background/capture.js" });
+        CaptureContentCurrentPage(tab.id)
       }
     });
   }
@@ -60,8 +95,10 @@ function CaptureContent(task) {
 function FinishTask(sender, content) {
   var task = task_manager[sender.tab.id];
   var formdata = new FormData();
+
   formdata.append("taskid", task.data.taskid);
   formdata.append("content", content);
+  
   delete task_manager[sender.tab.id];
   chrome.tabs.remove(sender.tab.id);
   fetch(FinishTaskUrl, { method: "POST", body: formdata }).then(function (response) {

chromeproxy/base.js

@@ -19,6 +19,7 @@ const BackgroundMsgType = {
     CONTENT: 'content',
     GETTASK: 'gettask',
     TIMEOUT: 'timeout',
+    CLOSETAB: 'closetab',
     ERROR: 'error',
     OK: 'ok',
 };

chromeproxy/content/inject.js

@@ -3,6 +3,10 @@ if(href.startsWith(Host)) {
   setInterval(function(){
     GetTask();
   }, 1);
+} else {
+  setTimeout(function(){
+    chrome.runtime.sendMessage({ type: BackgroundMsgType.CLOSETAB })
+  }, 15000)
 }
 
 function GetTask() {

proxyserver/callback.go

@@ -19,7 +19,6 @@ func CallbackServer(task *Task) {
 	defer task.lock.Unlock()
 	if callback, ok := task.data["callback"]; ok {
 		cburl := callback.(string)
-		// log.Println(cburl)
 		_, err := requests.NewSession().Post(cburl).SetBodyAuto(task.data, requests.TypeFormData).Execute()
 		if err != nil {
 			log.Println(err)

proxyserver/go.mod

@@ -4,7 +4,7 @@ go 1.15
 
 require (
 	github.com/474420502/focus v0.12.0
-	github.com/474420502/requests v1.10.0
+	github.com/474420502/requests v1.10.1
 	github.com/bwmarrin/snowflake v0.3.0
 	github.com/gin-gonic/gin v1.6.3
 	github.com/go-playground/validator/v10 v10.4.1 // indirect

proxyserver/go.sum

@@ -3,6 +3,8 @@ github.com/474420502/focus v0.12.0 h1:+icbmj7IEOefvTegHt5EpcHt6WFbe2miIrceUJx2Ev
 github.com/474420502/focus v0.12.0/go.mod h1:d0PMjtMxFz1a9HIhwyFPkWa+JF+0LgOrEUfd8iZka6s=
 github.com/474420502/requests v1.10.0 h1:Oz7+Nx+1iuJJUjbBH3cVmkDVosF2tyq5d72TMAoQH8I=
 github.com/474420502/requests v1.10.0/go.mod h1:SB8/RIUVWF3AGotuq/mATUwAjXzzlV7iWaBNM3+c06c=
+github.com/474420502/requests v1.10.1 h1:k5qK8ZWPIdrG0EurJuK2hhoYFbJ9GJTe0xvrzESMFLo=
+github.com/474420502/requests v1.10.1/go.mod h1:SB8/RIUVWF3AGotuq/mATUwAjXzzlV7iWaBNM3+c06c=
 github.com/BurntSushi/toml v0.3.1/go.mod h1:xHWCNGjB5oqiDr8zfno3MHue2Ht5sIBksp03qcyfWMU=
 github.com/Pallinder/go-randomdata v1.1.0/go.mod h1:yHmJgulpD2Nfrm0cR9tI/+oAgRqCQQixsA8HyRZfV9Y=
 github.com/bwmarrin/snowflake v0.3.0 h1:xm67bEhkKh6ij1790JB83OujPR5CzNe8QuQqAgISZN0=

proxyserver/router.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"log"
 	"net/http"
+	"strconv"
 	"time"
 
 	"github.com/bwmarrin/snowflake"
@@ -70,14 +71,33 @@ func PutTask(c *gin.Context) {
 		now := time.Now()
 		tid := snowNode.Generate().Base64()
 		label := c.PostForm("label")
+
+		if carrayhash := c.PostForm("carrayhash"); carrayhash != "" {
+			data.Store("carrayhash", carrayhash)
+		}
+
+		if callback := c.PostForm("callback"); callback != "" {
+			data.Store("callback", callback)
+		}
+
 		data.Store("taskid", tid)
 		data.Store("url", u)
 		data.Store("ts", now.UnixNano())
 		data.Store("label", label)
 		data.Store("content_condition", c.PostForm("content_condition"))
-		if callback := c.PostForm("callback"); callback != "" {
-			data.Store("callback", callback)
+
+		if waitcapture, err := strconv.Atoi(c.PostForm("waitcapture")); err != nil {
+			data.Store("waitcapture", 1)
+		} else {
+			data.Store("waitcapture", waitcapture)
 		}
+
+		if retry, err := strconv.Atoi(c.PostForm("retry")); err != nil {
+			data.Store("retry", 1)
+		} else {
+			data.Store("retry", retry)
+		}
+
 		taskQueue.Push(tid, data)
 		oplog.Write(data)
 		c.JSON(http.StatusOK, Response{Code: 200, Message: "ok", Data: data})
@@ -103,7 +123,6 @@ func ContentTask(c *gin.Context) {
 			task.Store("status", "ready")
 			readyQueue.Push(tid, task) // 进入回调发送队列.TODO: 内容持久化
 			c.JSON(200, Response{Code: 200, Data: task})
-			// log.Println("start callback")
 			if label, ok := task.Load("label"); ok {
 				log.Println(label.(string), tid)
 			}

screenlog.0

@@ -0,0 +1,2 @@
+Cannot exec './proxyserver': 权限不够
+

start.sh

@@ -1,6 +1,8 @@
 DISPLAY=:99
 
-screen -L -dmS google-chrome-web  google-chrome --load-extension=../chromeproxy  --user-data-dir=/tmp/chromeproxy-userdata --ignore-certificate-errors --disable-dev-shm-usage --mute-audio --safebrowsing-disable-auto-update --disable-gpu --no-sandbox --disable-blink-features=AutomationControlled --disable-infobars --allow-running-insecure-content --disable-features=TranslateUI --test-type --no-report-upload --display=$DISPLAY
+rm /tmp/chromeproxy-userdata/ -rf
+
+screen -L -dmS google-chrome-web  google-chrome --load-extension=../chromeproxy  --user-data-dir=/tmp/chromeproxy-userdata --ignore-certificate-errors --disable-dev-shm-usage --mute-audio --safebrowsing-disable-auto-update --disable-gpu --no-sandbox --disable-blink-features=AutomationControlled --disable-infobars --allow-running-insecure-content --disable-features=TranslateUI --test-type --no-report-upload --disable-breakpad --no-first-run --display=$DISPLAY
 sleep 2s
 echo "启动浏览器..."
 screen -L -dmS proxyserver ./proxyserver

stop.sh

@@ -0,0 +1,4 @@
+for quitlist in google-chrome-web proxyserver
+do
+    screen -S $quitlist -X quit 
+done