C# Selenium示例

引用包： Selenium.WebDriver

internal class Program
{
private static readonly TimeSpan DefaultTimeout = TimeSpan.FromSeconds(30);

    static void Main(string[] args)
     {
         Console.WriteLine("Please enter the path of the TSV file:");
         string inputPath = Console.ReadLine();
         if (!File.Exists(inputPath))
         {
             Console.WriteLine("The file does not exist!");
             Console.ReadKey();
             return;
         }
         string fileName = Path.GetFileNameWithoutExtension(inputPath);
         string directory = Path.GetDirectoryName(inputPath);
         string outputPath = $@"{directory}\{fileName}_output.tsv";
         var lines = new List<string>(File.ReadLines(inputPath));
         string baseLink = "https://bingdex.binginternal.com";

        using (IWebDriver driver = new EdgeDriver())
         {
             driver.Manage().Window.Maximize();
             WebDriverWait wait = new WebDriverWait(driver, DefaultTimeout);

            using (TextWriter writer = new StreamWriter(outputPath))
             {
                 //set header
                 writer.WriteLine(string.Join("\t", new string[] { "PageUrl", "Crawled_Status", "Crawled_Callisto_Status", "Crawled_Index_Status", "Served_Status", "Served_Reason", "Served_Warning", "Spam_Junk_List" }));

                for (int i = 0; i < lines.Count; i++)
                 {
                     var pageUrl = lines[i];
                     try
                     {
                         driver.Navigate().GoToUrl($"{baseLink}/home?url={pageUrl}");
                         Func<IWebDriver, bool> stageStatusCondition = x =>
                         {
                             var elements = x.FindElements(By.ClassName("stageStatus"));
                             return elements.Count == 4 && elements.All(element => !string.IsNullOrEmpty(element.Text));
                         };

                        WaitElementLoad(driver, stageStatusCondition);

                        var stageStatusList = driver.FindElements(By.ClassName("stageStatus"));
                         string crawled_status_text = stageStatusList[2].Text;
                         string crawled_status = ReplaceBlank(crawled_status_text);

                        string crawled_callisto_status = string.Empty;
                         string crawled_index_status = string.Empty;
                         string served_reason = string.Empty;
                         string served_warning = string.Empty;
                         string served_satus = ReplaceBlank(stageStatusList[3].Text);

                        if (IsElementPresent(driver, By.XPath("//th[contains(text(),'Callisto status')]")))
                         {
                             var callistostatusElement = driver.FindElement(By.XPath("//th[contains(text(),'Callisto status')]"));
                             string crawled_callisto_status_text = callistostatusElement.FindElement(By.XPath("following-sibling::td")).Text;
                             crawled_callisto_status = ReplaceBlank(crawled_callisto_status_text);
                         }

                        if (IsElementPresent(driver, By.XPath("//th[contains(text(),'Index status')]")))
                         {
                             var indexStatusElement = driver.FindElement(By.XPath("//th[contains(text(),'Index status')]"));
                             var crawled_index_status_text = indexStatusElement.FindElement(By.XPath("following-sibling::td")).Text;
                             crawled_index_status = ReplaceBlank(crawled_index_status_text);
                         }

                        if (IsElementPresent(driver, By.XPath("//th[contains(text(),'Reason')]")))
                         {
                             var reasonElement = driver.FindElement(By.XPath("//th[contains(text(),'Reason')]"));
                             var served_reason_text = reasonElement.FindElement(By.XPath("following-sibling::td")).Text;
                             served_reason = ReplaceBlank(served_reason_text);
                         }

                        if (IsElementPresent(driver, By.XPath("//th[contains(text(),'Warning')]")))
                         {
                             var warningElement = driver.FindElement(By.XPath("//th[contains(text(),'Warning')]"));
                             var served_warning_text = warningElement.FindElement(By.XPath("following-sibling::td")).Text;
                             served_warning = ReplaceBlank(served_warning_text);
                         }

                        var spamJunklistBtn = driver.FindElement(By.XPath("//*[@id=\"root\"]/div/div/div[1]/div/div[2]/button"));
                         spamJunklistBtn.Click();
                         Thread.Sleep(2000);

                        var spamJunkList = GetSpamJunkList(driver);
                         writer.WriteLine(string.Join("\t", new string[] { pageUrl, crawled_status, crawled_callisto_status, crawled_index_status, served_satus, served_reason, served_warning, spamJunkList }));
                     }
                     catch (Exception ex)
                     {
                         using (var errorWriter = new StreamWriter($@"{directory}\{fileName}_error.txt", true))
                         {
                             errorWriter.WriteLine(pageUrl);
                         }
                     }
                 }
             }
             Console.WriteLine("output file success!");
             Console.ReadKey();
         }
     }

    public static string ReplaceBlank(string str)
     {
         if (string.IsNullOrEmpty(str))
         {
             return str;
         }
         else
         {
             return str.Contains("\r") ? str.Substring(0, str.IndexOf("\r")) : str;
         }
     }

    public static bool IsElementPresent(IWebDriver driver, By by)
     {
         try
         {
             driver.FindElement(by);
             return true;
         }
         catch (NoSuchElementException)
         {
             return false;
         }
     }

     //Wait for element loading based on custom conditions
     public static void WaitElementLoad(IWebDriver driver, Func<IWebDriver, bool> condition)
     {
         try
         {
             WebDriverWait wait = new WebDriverWait(driver, DefaultTimeout);
             wait.Until(condition);
         }
         catch (WebDriverTimeoutException ex)
         {
             Console.WriteLine("Timeout waiting for elements to load inner text: " + ex.Message);
             throw ex;
         }
     }

    public static string GetSpamJunkList(IWebDriver driver)
     {
         string spamJunkList = string.Empty;
         List<SpamJunk> spamJunks = new List<SpamJunk>();
         try
         {
             var rows = driver.FindElements(By.ClassName("ms-List-cell"));
             if (rows.Count > 0)
             {
                 foreach (var item in rows)
                 {
                     if (IsElementPresent(driver, By.ClassName("ms-DetailsRow-cell")))
                     {
                         var cells = item.FindElements(By.ClassName("ms-DetailsRow-cell"));
                         if (cells.Count == 5)
                         {
                             spamJunks.Add(new SpamJunk()
                             {
                                 Group = cells?[0]?.Text,
                                 Lable = cells?[1]?.Text,
                                 Signature = cells?[2]?.Text,
                                 Source = cells?[3]?.Text,
                                 Tier = cells?[4]?.Text
                             });
                         }
                         continue;
                     }

                }
                 spamJunkList = JsonConvert.SerializeObject(spamJunks);
             }
             else
             {
                 spamJunkList = "No spam data found";
             }
         }
         catch (NoSuchElementException)
         {

            spamJunkList = "No spam data found";
         }
         return spamJunkList;
     }

}

class SpamJunk
 {
     public string Group { get; set; }

    public string Lable { get; set; }

    public string Signature { get; set; }

    public string Source { get; set; }

    public string Tier { get; set; }
 }