WebView2 範例, Regex, 網頁爬文

webview2.sample.cs

private async void Form1_Load(object sender, EventArgs e)
{
  // init
  cboStatus.SelectedIndex = 0;

  // init WebView2 → webAgent
  await webAgent.EnsureCoreWebView2Async(null);
  webAgent.CoreWebView2.Navigate("https://www.itfseafarers.org/en/abandonment-list/seafarer-abandonment");
}

private async void button2_Click(object sender, EventArgs e)
{
  labelResult.Text = string.Empty;
  progressBar.Value = 0;

  //string status = "disputed"; // All | open | disputed | inactive | resolved 。
  string status = (string)(cboStatus.Text ?? "ALL");

  string queryJs = $"document.querySelector('#edit-field-status-list-value').value = '{status}'; "
    + @"document.querySelector('#views-exposed-form-itf-current-cases-block-1').submit();";
  await webAgent.CoreWebView2.ExecuteScriptAsync(queryJs);

  progressBar.Value = 10;

  // 讓子彈飛
  await Task.Delay(1000);
  progressBar.Value = 20;
  await Task.Delay(1000);
  progressBar.Value = 30;
  await Task.Delay(1000);
  progressBar.Value = 40;
  await Task.Delay(1000);
  progressBar.Value = 50;
  await Task.Delay(1000);
  progressBar.Value = 60;

  // 檢畫面確定status 查詢條件無誤!
  string regionResult = await webAgent.CoreWebView2.ExecuteScriptAsync(@"document.querySelector('#edit-field-region-target-id').value");
  string statusResult = await webAgent.CoreWebView2.ExecuteScriptAsync(@"document.querySelector('#edit-field-status-list-value').value");
  labelResult.Text = $"{regionResult} | {statusResult}";

  progressBar.Value = 70;

  // 取回資訊 html 或直接用 js 取值出來
  string html = await webAgent.CoreWebView2.ExecuteScriptAsync(@"document.querySelector('#block-views-block-itf-current-cases-block-1 > div > div > table > tbody').outerHTML");
  string decodedHtml = Regex.Unescape(html); // 解碼:Unicode, \u003C => `<`
  string decodedHtml2 = WebUtility.HtmlDecode(decodedHtml); // 解碼:htmle 

  progressBar.Value = 80;

  // 解析資訊 => datainfo with RegEx 
  const string pattern =
         @"(<td .*-vessel-name"">(?<vesselName>.*)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-imo"">(?<imo>.*)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-flag"">(?<flag>.*)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-port"">(?<port>.*)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-country"">(?<country>.*)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-owed-wages"">(?<owedWages>.*)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-seafarers"">(?<seafarers>.*)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-nationalities"">(?<nationalities>.*)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-reported-to-itf is-active"">(?<reportedToItf>[\s\S]*?)<\/td>)[\s\r\n\t]*" +
         @"(<td .*-status-list"">(?<statusList>.*)<\/td>)";

  List<SeafarerAbandonment> infoList = new();
  foreach (Match tr in Regex.Matches(decodedHtml2, @"<tr>([\s\S\r\n\t]*?)<\/tr>"))
  {
    Match match = Regex.Match(tr.Value, pattern);

    if (match.Success)
    {
      var info = new SeafarerAbandonment();
      info.VesselName = match.Groups["vesselName"].ToString().Trim();
      info.Imo = match.Groups["imo"].ToString().Trim();
      info.Flag = match.Groups["flag"].ToString().Trim();
      info.Port = match.Groups["port"].ToString().Trim();
      info.Country = match.Groups["country"].ToString().Trim();
      info.OwedWages = match.Groups["owedWages"].ToString().Trim();
      info.Seafarers = match.Groups["seafarers"].ToString().Trim();
      info.Nationalities = match.Groups["nationalities"].ToString().Trim();
      info.ReportedToItf = match.Groups["reportedToItf"].ToString().Trim();
      info.StatusList = match.Groups["statusList"].ToString().Trim();

      infoList.Add(info);
    }
  }

  progressBar.Value = 90;

  //## 輸出到畫面
  string dataJson = JsonSerializer.Serialize(infoList, new JsonSerializerOptions
  {
    WriteIndented = true,
    Encoder = System.Text.Encodings.Web.JavaScriptEncoder.UnsafeRelaxedJsonEscaping, // 中文字不編碼
  });

  textBox1.Text = dataJson;

  progressBar.Value = 100;
}