private void cmbTitleXpath_SelectionChanged(object sender, SelectionChangedEventArgs e)
{
var senderCombox = ((ComboBox)sender).SelectedValue;
if (iscmbLoading || senderCombox == null || (selectedPattern.Path.TitleXPath != null && selectedPattern.Path.TitleXPath.Equals(senderCombox.ToString())))
{
return;
}
selectedPattern.Path.TitleXPath = senderCombox.ToString();
var contents = XpathParser.ParseList(httpResult.Content, JsonConvert.SerializeObject(selectedPattern.Path), channelUrl);
selectedPattern.Contents = contents.Articles.ToArray();
ChangeSelectedXpath();
}
private void btnOutPut_Click(object sender, RoutedEventArgs e)
{
try
{
var saveFileDialog1 = new SaveFileDialog
{
Filter = "Excel|*.xlsx|Excel 2003|*.xls",
Title = "Save File"
};
saveFileDialog1.ShowDialog();
if (string.IsNullOrEmpty(saveFileDialog1.FileName))
{
return;
}
if (selectedPattern != null && selectedPattern.Path != null)
{
var contents = XpathParser.ParseList(httpResult.Content, JsonConvert.SerializeObject(selectedPattern.Path), channelUrl);
List <Article> articles = new List <Article>();
foreach (var article in contents.Articles)
{
if (!string.IsNullOrWhiteSpace(article.Author) && article.Author.Contains(" "))
{
article.Author = article.Author.Split(' ')[0];
}
//Article _article = ThrinaxHelper.GetArticleAndFormatter(article.Url);
//article.Content = _article.Content;
//article.HtmlContent = _article.HtmlContent;
//if (article.PubDate == null || article.PubDate <= DateTime.Parse("1970-01-01"))
//{
// article.PubDate = _article.PubDate;
//}
articles.Add(article);
}
ExcelHelper.SaveExcelSheet(articles, saveFileDialog1.FileName, "Thrinax");
}
}
catch (Exception ex)
{
//log the err.
}
MessageBox.Show("Finish");
}
private void ChangeSelectedXpath()
{
iscmbLoading = true;
btnRight.IsEnabled = false;
btnWrong.IsEnabled = false;
cmbUrlXpath.IsEnabled = false;
cmbTitleXpath.IsEnabled = false;
cmbTimeXpath.IsEnabled = false;
cmbAuthorXpath.IsEnabled = false;
cmbUrlXpath.Items.Clear();
cmbTitleXpath.Items.Clear();
cmbTimeXpath.Items.Clear();
cmbAuthorXpath.Items.Clear();
cmbUrlXpath.Items.Add(selectedPattern.Path.UrlXPath);
//在 listPatterns 不为 null 的情况下绑定第一个模式的数据
cmbTitleXpath.Items.Add(selectedPattern.Path.TitleXPath);
cmbTimeXpath.Items.Add(selectedPattern.Path.DateXPath ?? "");
//获取备选的时间模式
if (selectedPattern.BackUpPaths.ContainsKey(PatternType.Date))
{
List <string> tempPaths = selectedPattern.BackUpPaths[PatternType.Date];
//判断是否为空，以及是否和当选的Xpath相同
if (tempPaths != null)
{
foreach (string tempPath in tempPaths)
{
if (!string.IsNullOrEmpty(tempPath) && !tempPath.Equals(selectedPattern.Path.DateXPath))
{
cmbTimeXpath.Items.Add(tempPath);
}
}
}
}
if (!string.IsNullOrEmpty(selectedPattern.Path.DateXPath))
{
cmbTimeXpath.Items.Add("");
}
cmbAuthorXpath.Items.Add(selectedPattern.Path.AuthorXPath ?? "");
//获取备选的作者模式
if (selectedPattern.BackUpPaths.ContainsKey(PatternType.Author))
{
List <string> tempPaths = selectedPattern.BackUpPaths[PatternType.Author];
//判断是否为空，以及是否和当选的Xpath相同
if (tempPaths != null)
{
foreach (string tempPath in tempPaths)
{
if (!string.IsNullOrEmpty(tempPath) && !tempPath.Equals(selectedPattern.Path.AuthorXPath))
{
cmbAuthorXpath.Items.Add(tempPath);
}
}
}
}
if (!string.IsNullOrEmpty(selectedPattern.Path.AuthorXPath))
{
cmbAuthorXpath.Items.Add("");
}
cmbUrlXpath.SelectedIndex = 0;
cmbTitleXpath.SelectedIndex = 0;
cmbTimeXpath.SelectedIndex = 0;
cmbAuthorXpath.SelectedIndex = 0;
var contents = XpathParser.ParseList(httpResult.Content, JsonConvert.SerializeObject(selectedPattern.Path), channelUrl);
//绑定出事DataGrid
dataGridItems.ItemsSource = contents.Articles;
btnRight.IsEnabled = true;
btnWrong.IsEnabled = true;
cmbTitleXpath.IsEnabled = true;
cmbTimeXpath.IsEnabled = true;
cmbAuthorXpath.IsEnabled = true;
iscmbLoading = false;
}
private void RefreshUI()
{
selectedPattern = null;
if (string.IsNullOrEmpty(channelUrl) || listPagePatterns == null)
{
this.Dispatcher.Invoke(new Action(() => {
//控制浏览区与判断区样式
geckoFormHost.IsEnabled = false;
geckoWebBrowser.Navigate("about:blank");
geckoWebBrowser.IsAccessible = false;
btnRight.IsEnabled = false;
btnWrong.IsEnabled = false;
//清除表格和下拉框样式与数据
dataGridItems.IsEnabled = false;
cmbUrlXpath.IsEnabled = false;
cmbTitleXpath.IsEnabled = false;
cmbTimeXpath.IsEnabled = false;
cmbAuthorXpath.IsEnabled = false;
cmbUrlXpath.Items.Clear();
cmbTitleXpath.Items.Clear();
cmbTimeXpath.Items.Clear();
cmbAuthorXpath.Items.Clear();
//清除模式选择按钮状态
btnModel1.IsChecked = false;
btnModel2.IsChecked = false;
btnModel3.IsChecked = false;
btnModel4.IsChecked = false;
btnModel5.IsChecked = false;
btnModel1.IsEnabled = false;
btnModel2.IsEnabled = false;
btnModel3.IsEnabled = false;
btnModel4.IsEnabled = false;
btnModel5.IsEnabled = false;
}));
}
else
{
this.Dispatcher.Invoke(new Action(() =>
{
//启用浏览区域与操作区域，加载缓存的网页
geckoFormHost.IsEnabled = true;
try
{
if (geckoWebBrowser.Url != new Uri(channelUrl))
{
geckoWebBrowser.Navigate(channelUrl);
}
}
catch { }
geckoWebBrowser.IsAccessible = true;
btnRight.IsEnabled = true;
btnWrong.IsEnabled = true;
cmbUrlXpath.Items.Clear();
cmbTitleXpath.Items.Clear();
cmbTimeXpath.Items.Clear();
cmbAuthorXpath.Items.Clear();
//在 listPatterns 不为 null 的情况下绑定第一个模式的数据
if (listPagePatterns != null && listPagePatterns.Count > 0)
{
selectedPattern = listPagePatterns[0];
cmbUrlXpath.Items.Add(selectedPattern.Path.UrlXPath);
cmbTitleXpath.Items.Add(selectedPattern.Path.TitleXPath);
cmbTimeXpath.Items.Add(selectedPattern.Path.DateXPath ?? "");
//获取备选的时间模式
if (selectedPattern.BackUpPaths.ContainsKey(PatternType.Date))
{
List <string> tempPaths = selectedPattern.BackUpPaths[PatternType.Date];
//判断是否为空，以及是否和当选的Xpath相同
if (tempPaths != null)
{
foreach (string tempPath in tempPaths)
{
if (!string.IsNullOrEmpty(tempPath) && !tempPath.Equals(selectedPattern.Path.DateXPath))
{
cmbTimeXpath.Items.Add(tempPath);
}
}
}
}
if (!string.IsNullOrEmpty(selectedPattern.Path.DateXPath))
{
cmbTimeXpath.Items.Add("");
}
cmbAuthorXpath.Items.Add(selectedPattern.Path.AuthorXPath ?? "");
//获取备选的作者模式
if (selectedPattern.BackUpPaths.ContainsKey(PatternType.Author))
{
List <string> tempPaths = selectedPattern.BackUpPaths[PatternType.Author];
//判断是否为空，以及是否和当选的Xpath相同
if (tempPaths != null)
{
foreach (string tempPath in tempPaths)
{
if (!string.IsNullOrEmpty(tempPath) && !tempPath.Equals(selectedPattern.Path.AuthorXPath))
{
cmbAuthorXpath.Items.Add(tempPath);
}
}
}
}
if (!string.IsNullOrEmpty(selectedPattern.Path.AuthorXPath))
{
cmbAuthorXpath.Items.Add("");
}
cmbUrlXpath.SelectedIndex = 0;
cmbTitleXpath.SelectedIndex = 0;
cmbTimeXpath.SelectedIndex = 0;
cmbAuthorXpath.SelectedIndex = 0;
txtBaseXpath.Text = selectedPattern.Path?.ItemRootXPath;
var contents = XpathParser.ParseList(httpResult.Content, JsonConvert.SerializeObject(selectedPattern.Path), channelUrl);
//绑定出事DataGrid
dataGridItems.ItemsSource = contents.Articles;
}
//启用表格和下拉框
dataGridItems.IsEnabled = true;
cmbUrlXpath.IsEnabled = false;
cmbTitleXpath.IsEnabled = true;
cmbTimeXpath.IsEnabled = true;
cmbAuthorXpath.IsEnabled = true;
//根据数据情况确定模式按钮的可用性
btnModel1.IsEnabled = listPagePatterns.Count > 0;
btnModel2.IsEnabled = listPagePatterns.Count > 1;
btnModel3.IsEnabled = listPagePatterns.Count > 2;
btnModel4.IsEnabled = listPagePatterns.Count > 3;
btnModel5.IsEnabled = listPagePatterns.Count > 4;
btnModel1.IsChecked = true;
btnModel2.IsChecked = false;
btnModel3.IsChecked = false;
btnModel4.IsChecked = false;
btnModel5.IsChecked = false;
ChangeDisplayGecko(selectedPattern);
}));
}
}
/// <summary>
/// 验证一个List的模式是否能应用于某一个页面（只是检查是否明显不可能）
/// </summary>
/// <param name="Url"></param>
/// <param name="RootNode"></param>
/// <param name="XPath"></param>
/// <returns></returns>
public bool ValidateListXPath(string Url, HtmlNode RootNode, XpathPattern XPath)
{
if (string.IsNullOrEmpty(Url) || RootNode == null || XPath == null)
{
return(false);
}
List <Article> Content = XpathParser.ExtractItemFromList(Url, RootNode, XPath);
if (Content == null || Content.Count < 3)
{
return(false);
}
int TitleCount = 0, DateCount = 0, ViewCount = 0, ReplyCount = 0, MediaCount = 0, AuthorCount = 0;
foreach (Article ele in Content)
{
if (!string.IsNullOrEmpty(ele.Title) && !string.IsNullOrEmpty(ele.Url))
{
TitleCount++;
}
if (!string.IsNullOrEmpty(XPath.DateXPath) && ele.PubDate != null)
{
DateCount++;
}
if (!string.IsNullOrEmpty(XPath.ViewXPath) && ele.ViewDataList?.FirstOrDefault()?.View >= 0)
{
ViewCount++;
}
if (!string.IsNullOrEmpty(XPath.ReplyXPath) && ele.ViewDataList?.FirstOrDefault()?.Reply >= 0)
{
ReplyCount++;
}
if (!string.IsNullOrEmpty(XPath.MediaNameXPath) && !string.IsNullOrEmpty(ele.MediaName))
{
MediaCount++;
}
if (!string.IsNullOrEmpty(XPath.AuthorXPath) && !string.IsNullOrEmpty(ele.Author))
{
AuthorCount++;
}
}
if (TitleCount < Content.Count * 0.9)
{
return(false);
}
if (!string.IsNullOrEmpty(XPath.DateXPath) && DateCount < Content.Count * 0.9)
{
return(false);
}
if (!string.IsNullOrEmpty(XPath.ViewXPath) && ViewCount < Content.Count * 0.4)
{
return(false);
}
if (!string.IsNullOrEmpty(XPath.ReplyXPath) && ReplyCount < Content.Count * 0.1)
{
return(false);
}
if (!string.IsNullOrEmpty(XPath.MediaNameXPath) && MediaCount < Content.Count * 0.9)
{
return(false);
}
if (!string.IsNullOrEmpty(XPath.AuthorXPath) && AuthorCount < Content.Count * 0.9)
{
return(false);
}
return(true);
}
