V0.3.2

ElectronJS/main.js

@@ -296,6 +296,7 @@ async function runBrowser(lang = "en", user_data_folder = '') {
     }
     options.addExtensions(path.join(__dirname, "XPathHelper.crx"));
     options.setChromeBinaryPath(chromeBinaryPath);
+    options.add
     if (user_data_folder != "") {
         let dir = path.join(task_server.getDir(), user_data_folder);
         console.log(dir);

ElectronJS/tasks/96.json

@@ -0,0 +1 @@
+{"id":96,"name":"[2305.07067] SigRec: Automatic Recovery of Function Signatures in Smart Contracts","url":"https://arxiv.org/abs/2305.07067","links":"https://arxiv.org/pdf/2008.03554.pdf","create_time":"6/2/2023, 1:00:27 AM","version":"0.3.2","containJudge":false,"desc":"https://arxiv.org/abs/2305.07067","inputParameters":[{"id":0,"name":"urlList_0","nodeId":1,"nodeName":"打开网页","value":"https://arxiv.org/pdf/2008.03554.pdf","desc":"要采集的网址列表，多行以\\n分开","type":"string","exampleValue":"https://arxiv.org/pdf/2008.03554.pdf"}],"outputParameters":[],"graph":[{"index":0,"id":0,"parentId":0,"type":-1,"option":0,"title":"root","sequence":[1,2],"parameters":{"history":1,"tabIndex":0,"useLoop":false,"xpath":"","wait":0},"isInLoop":false},{"id":1,"index":1,"parentId":0,"type":0,"option":1,"title":"打开网页","sequence":[],"isInLoop":false,"position":0,"parameters":{"useLoop":false,"xpath":"","wait":0,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"url":"https://arxiv.org/abs/2305.07067","links":"https://arxiv.org/pdf/2008.03554.pdf","maxWaitTime":10,"scrollType":0,"scrollCount":1,"scrollWaitTime":1}},{"id":2,"index":2,"parentId":0,"type":0,"option":2,"title":"点击元素","sequence":[],"isInLoop":false,"position":1,"parameters":{"history":4,"tabIndex":-1,"useLoop":false,"xpath":"//*[contains(@class, \"download-pdf\")]","wait":2,"beforeJS":"","beforeJSWaitTime":0,"afterJS":"","afterJSWaitTime":0,"scrollType":0,"scrollCount":1,"scrollWaitTime":1,"maxWaitTime":10,"paras":[],"allXPaths":["/html/body/div[2]/main[1]/div[1]/div[1]/div[2]/div[1]/ul[1]/li[1]/a[1]","//a[contains(., 'PDF')]","//A[@class='abs-button download-pdf']"]}}]}

ExecuteStage/.vscode/launch.json

@@ -12,7 +12,7 @@
             "console": "integratedTerminal",
             "justMyCode": true,
             // "args": ["--id", "38", "--read_type", "local", "--headless", "1"]
-            "args": ["--id", "[5]", "--headless", "0", "--user_data", "0"]
+            "args": ["--id", "[80]", "--headless", "0", "--user_data", "0"]
         }
     ]
 }

ExecuteStage/easyspider_executestage.py

@@ -702,10 +702,18 @@ class BrowserThread(Thread):
             textbox.send_keys(Keys.SHIFT, Keys.END)
             # Send the DELETE key
             textbox.send_keys(Keys.DELETE)
+            value = ""
             if para["useLoop"]:
-                textbox.send_keys(loopValue)
+                value = loopValue
             else:
-                textbox.send_keys(para["value"])
+                value = para["value"]
+            pattern = r'Field\["([^"]+)"\]' # 将value中的Field[""]替换为outputParameters中的键值
+            try:
+                replaced_text = re.sub(pattern, lambda match: self.outputParameters.get(match.group(1), ''), value)
+            except:
+                replaced_text = value
+            value = replaced_text
+            textbox.send_keys(value)
             self.execute_code(2, para["afterJS"], para["afterJSWaitTime"], textbox) # 执行后置js
             # global bodyText  # 每次执行点击，输入元素和打开网页操作后，需要更新bodyText
             self.bodyText = self.browser.find_element(By.CSS_SELECTOR, "body").text
@@ -1089,6 +1097,22 @@ if __name__ == '__main__':
     option.add_argument(
         "--disable-blink-features=AutomationControlled")  # TMALL 反扒
     options.add_argument("--disable-blink-features=AutomationControlled")  # TMALL 反扒
+    options.add_experimental_option("prefs", {
+            "download.default_directory": "Data/",  # 设置文件下载路径
+            "download.prompt_for_download": False,  # 禁止下载提示框
+            "plugins.plugins_list": [{"enabled": False, "name": "Chrome PDF Viewer"}],
+            "download.directory_upgrade": True,
+            "download.extensions_to_open": "applications/pdf",
+            "plugins.always_open_pdf_externally": True  # 总是在外部程序中打开PDF
+        })
+    option.add_experimental_option("prefs", {
+            "download.default_directory": "Data/",  # 设置文件下载路径
+            "download.prompt_for_download": False,  # 禁止下载提示框
+            "plugins.plugins_list": [{"enabled": False, "name": "Chrome PDF Viewer"}],
+            "download.directory_upgrade": True,
+            "download.extensions_to_open": "applications/pdf",
+            "plugins.always_open_pdf_externally": True  # 总是在外部程序中打开PDF
+        })
     print(options)
     threads = []
     for i in c.id: