stable

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("content", content);
+} else {
+    Tell("notwant", content);
+}
+

chromeproxy/background/config.js

@@ -2,22 +2,40 @@
 chrome.webRequest.onBeforeRequest.addListener(function (details) {
     if (details.url.startsWith("http://eson.config")) {
       var params = new URL(details.url).searchParams;
-      UpdateHost(params.get("taskurl"))
-      chrome.tabs.remove(details.tabId);
-      chrome.tabs.query({ currentWindow: true }, function (tabs) {
-        var count = 0
-        for (var tab of tabs) {
-          if (!tab.url.startsWith("chrome://")) {
-            count++;
-            chrome.tabs.executeScript(tab.id, { runAt: "document_end", code: `window.location.href = "${Host}"` })
-          } else {
-            chrome.tabs.remove(tab.id);
-          }
+
+      for( key of params.keys()) {
+        switch(key) {
+          case "proxyserver":
+            ProxyServer = params.get(key)
+          break
         }
-        for (var i = count; i < WorkerNumber; i++) {
-          chrome.tabs.create({ url: `${Host}` });
-        }
-      })
+      }
+
+      var taskurl = params.get("taskurl")
+      if(taskurl !== "") {
+        UpdateHost(taskurl)
+        removeConfigTabAndCreateGetTaskTab(details);
+      }
+
       return { cancel: true };
     }
-  }, { 'urls': ["<all_urls>"] }, ['blocking']);
+  }, { 'urls': ["<all_urls>"] }, ['blocking']);
+
+
+function removeConfigTabAndCreateGetTaskTab(details) {
+  chrome.tabs.remove(details.tabId);
+  chrome.tabs.query({ currentWindow: true }, function (tabs) {
+    var count = 0
+    for (var tab of tabs) {
+      if (!tab.url.startsWith("chrome://")) {
+        count++;
+        chrome.tabs.executeScript(tab.id, { runAt: "document_end", code: `window.location.href = "${Host}"` })
+      } else {
+        chrome.tabs.remove(tab.id);
+      }
+    }
+    for (var i = count; i < WorkerNumber; i++) {
+      chrome.tabs.create({ url: `${Host}` });
+    }
+  })
+}

chromeproxy/background/proxy.js

@@ -0,0 +1,21 @@
+
+
+var config = {
+    mode: "pac_script",
+    pacScript: {
+        data: 
+        `
+        ProxyServer = ${ProxyServer};
+        function FindProxyForURL(url, host) {
+            if(ProxyServer != undefined) {
+                return ProxyServer;
+            }
+            return 'DIRECT';
+        }`
+    }
+};
+
+
+chrome.proxy.settings.set(
+    { value: config, scope: 'regular' },
+    function () { });

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 +60,31 @@ function GetTask(sender) {
   })
 }
 
+function CaptureContentCurrentPage(tabid) {
+  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) {
-        chrome.tabs.executeScript(tab.id, { runAt: "document_end", code: condition }, function () {
-          chrome.tabs.executeScript(tab.id, { runAt: "document_end", file: "background/capture.js" });
+
+      if (condition != undefined) {
+        chrome.tabs.executeScript(tab.id, { runAt: "document_start", code: condition }, function () {
+          CaptureContentCurrentPage(tab.id)
         });
       } else {
-        chrome.tabs.executeScript(tab.id, { runAt: "document_end", file: "background/capture.js" });
+        CaptureContentCurrentPage(tab.id)
       }
     });
   }
@@ -60,8 +93,10 @@ 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

@@ -1,17 +1,19 @@
-
 var Host = "http://localhost:7123"
 
 var GetTaskUrl = `${Host}/task/get`;
 var FinishTaskUrl = `${Host}/task/content`;
 var ErrorTaskUrl = `${Host}/task/error`;
+var ConfigUrl = `${Host}/task/config`
 
 var WorkerNumber = 1;
+var ProxyServer = undefined; 
 
 function UpdateHost(host) {
     Host = host 
     GetTaskUrl = `${Host}/task/get`
     FinishTaskUrl = `${Host}/task/content`
     ErrorTaskUrl = `${Host}/task/error`
+    ConfigUrl = `${Host}/task/config`
 }
 
 const BackgroundMsgType = {
@@ -19,6 +21,7 @@ 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() {

chromeproxy/manifest.json

@@ -39,7 +39,7 @@
    "key": "MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAlurgt0Ykv740tjk1ebeiSX6UVVRRhxVvh+FvRMTe9PKzkAKUDEW2IjNUz1swxZA8ILnMvtyamAfiErTOxUFB8+zCADU/CK2YEhqUJo3tmHCg6EP2XJL220ZXjmIeft1AqJV1BmGRLhm8VnH8dJ2EThfcflx4JEdlh0/aLJr6UVjF2hPVX8JLAMTVjEfiC82KDNGgXDADqBm3E/6n+Dx+3KhgIYTXKIMk+qRVaOhjAJLh8a9OrDBwpylP5RDifTAyVVa9UOyoLNqynzC9oLabWUr1ovWAOiivhYknFCXnl5q971iNSFpmjc+ZW+aK+TRjMnJF84IeA170corVG3KgnwIDAQAB",
    "manifest_version": 2,
    "name": "ChromeProxy",
-   "permissions": ["background", "http://*/*", "https://*/*",  "storage","tabs" , "activeTab", "webRequest", "webRequestBlocking", "<all_urls>", "webNavigation"],
+   "permissions": ["background", "http://*/*", "https://*/*", "proxy", "storage","tabs" , "activeTab", "webRequest", "webRequestBlocking", "<all_urls>", "webNavigation"],
    "short_name": "ChromeProxy",
    "version": "0.1.0"
 }

chromeproxy/pac.js

@@ -0,0 +1 @@
+

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/config.go

@@ -7,7 +7,7 @@ import (
 	"gopkg.in/yaml.v2"
 )
 
-var config = &Config{}
+var config = &LocalConfig{}
 
 func init() {
 	f, err := os.Open("config.yaml")
@@ -21,8 +21,8 @@ func init() {
 	}
 }
 
-// Config 本地yaml配置对象
-type Config struct {
+// LocalConfig 本地yaml配置对象
+type LocalConfig struct {
 	Server struct {
 		URI string `yaml:"uri"`
 	} `yaml:"server"`
@@ -41,3 +41,8 @@ func initChromeProxy() {
 		panic(err)
 	}
 }
+
+// ChromeConfig chrome的配置, 包括代理
+type ChromeConfig struct {
+	PacScript string
+}

proxyserver/go.mod

@@ -3,8 +3,8 @@ module git.nonolive.co/eson.hsm/proxyserver
 go 1.15
 
 require (
-	github.com/474420502/focus v0.12.0
-	github.com/474420502/requests v1.10.0
+	github.com/474420502/focus v0.13.1
+	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

@@ -1,8 +1,14 @@
 cloud.google.com/go v0.26.0/go.mod h1:aQUYkXzVsufM+DwF1aE+0xfcU+56JwCaLick0ClmMTw=
 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/focus v0.13.0 h1:68G0txM38EinNitcfpP0752k8tF4JRJ0r1mmdH6FoGc=
+github.com/474420502/focus v0.13.0/go.mod h1:d0PMjtMxFz1a9HIhwyFPkWa+JF+0LgOrEUfd8iZka6s=
+github.com/474420502/focus v0.13.1 h1:HwJCqY60eqRAkwKtIefJib1ofBfcvVh5ZTU2xE8Gmkk=
+github.com/474420502/focus v0.13.1/go.mod h1:SrqNq63qSx53TkvVWOuHTbTSvAQTNO8ZUrufXm0Ncq8=
 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=
@@ -125,6 +131,7 @@ golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7w
 golang.org/x/term v0.0.0-20201117132131-f5c789dd3221/go.mod h1:Nr5EML6q2oocZ2LXRh80K7BxOlk5/8JxuGnuhpl+muw=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
+golang.org/x/text v0.3.4/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 golang.org/x/tools v0.0.0-20190114222345-bf090417da8b/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 golang.org/x/tools v0.0.0-20190226205152-f727befe758c/go.mod h1:9Yl7xja0Znq3iFh3HoIrodX9oNMXvdceNzlUR8zjMvY=

proxyserver/router.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"log"
 	"net/http"
+	"strconv"
 	"time"
 
 	"github.com/bwmarrin/snowflake"
@@ -41,10 +42,9 @@ func init() {
 	task := engine.Group("/task")
 	task.GET("/get", GetTask)
 	task.POST("/put", PutTask)
-
 	task.POST("/content", ContentTask)
-
 	task.POST("/error", ErrorTask)
+	task.GET("/config", GetConfig)
 
 	task.GET("/ready", ReadyTask)
 	task.POST("/ack", AckTask)
@@ -70,14 +70,33 @@ 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 +122,6 @@ 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)
 			}
@@ -155,6 +173,11 @@ func ReadyTask(c *gin.Context) {
 	c.JSON(http.StatusOK, Response{Code: 404, Message: fmt.Sprintf("taskid: %s is not found", tid)})
 }
 
+// GetConfig 获取任务配置
+func GetConfig(c *gin.Context) {
+
+}
+
 // ErrorTask 任务错误无法完成
 func ErrorTask(c *gin.Context) {
 	tid := c.PostForm("taskid")

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