private async void ExtractImageWithHtmlAgilityPack(string html)
{
try
{
string value = "";
var imageList = await HtmlAgilityPackUtil.GetImgFromHtmlAsync(html);
for (int i = 0; i < imageList.Count; i++)
{
value = imageList[i];
if (value.StartsWith("//"))
{
value = "http:" + value;
}
if (value.Contains(":") == false)
{
value = BaseUrl + value;
}
AddToCollection(new UrlStruct()
{
Id = i + 1, Status = "", Title = "", Url = value
});
}
ShowStatusText($"已抓取到{imageCollection.Count}个图像");
}
catch (Exception ex)
{
ShowStatusText(ex.Message);
}
}
/// <summary>
/// XPath 是一门在 XML 文档中查找信息的语言。XPath 用于在 XML 文档中通过元素和属性进行导航。
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
private void btn_XPathQuery_Click(object sender, RoutedEventArgs e)
{
TextRange tr = new TextRange(rbox_XPathInput.Document.ContentStart, rbox_XPathInput.Document.ContentEnd);
var html = tr.Text;
if (string.IsNullOrEmpty(html))
{
EMessageBox.Show("请输入html");
return;
}
var result = HtmlAgilityPackUtil.XPathQuery(html, this.tbox_XPath.Text.Trim());
if (result != null)
{
Paragraph paragraph = new Paragraph();
foreach (var item in result)
{
paragraph.Inlines.Add(new Run(item.OuterHtml + Environment.NewLine));
}
this.rbox_XPathOutput.Document = new FlowDocument(paragraph);
}
else
{
this.rbox_XPathOutput.Document.Blocks.Clear();
}
}
private async Task <List <TagImg> > SearchBingImage(string keyword, int page = 1)
{
List <TagImg> searchImgList = new List <TagImg>();
var start = 1;
if (page > 1)
{
start = page * PageImageNum + 1;
}
var url = UrlUtil.CNBingImageDetailUrl.Replace("[keyword]", keyword).Replace("[start]", start.ToString());
searchImgList = await HtmlAgilityPackUtil.GetBingImgFromUrlAsync(url);
return(searchImgList);
}
private async void ExtractBackgroundImage(object html)
{
//我这里是写的div，可能页面上用来显示图片的不一定是div，是其它元素也说不定，如li ol ul
var xpath = "//div";
var result = HtmlAgilityPackUtil.XPathQuery(html.ToString(), xpath);
foreach (var item in result)
{
var classAttribute = item.Attributes["class"];
if (classAttribute == null)
{
continue;
}
var className = classAttribute.Value;
var script = $"getComputedStyle(document.getElementsByClassName('{className}')[0]).backgroundImage";
//执行js
var backgroundImage = await globalData.Browser.browser.EvaluateScriptAsync(script);
if (backgroundImage.Result != null && backgroundImage.Result.ToString() != "none")
{
var mathch = RegexUtil.RegexMatch(backgroundImage.Result.ToString(), RegexPattern.MatchImgPattern);
if (mathch.Success)
{
lock (obj)
{
Dispatcher.Invoke(() =>
{
backgroundImageList.Add(mathch.Value);
ShowStatusText($"已抓取到{backgroundImageList.Count}个图像");
});
}
}
}
}
if (backgroundImageList.Count == 0)
{
ShowStatusText("解析已完成，未抓取到任何图像");
}
}
private async void StartScroll(string html)
{
//第一次抓取内容完成，开始滚动页面
//获取高度 document.body.clientHeight
var getHeightJs = "document.body.clientHeight";
//用js控制滚动
//这里也可以直接用Selenium去驱动浏览器滚动
var scrollJs = "window.scroll(0,{0})";
var height = await globalData.Browser.EvaluateJavaScriptAsync(getHeightJs);
//无限循环滚动
while (true)
{
globalData.Browser.ExecuteJavaScript(string.Format(scrollJs, height));
var oldHeight = height;
height = await globalData.Browser.EvaluateJavaScriptAsync(getHeightJs);
if (height == oldHeight)
{
break;
}
//todo 登录操作
//使用js填入登录框内容 模拟点击登录
//由于这里仅做示例不针对任何网站
await Task.Delay(1000);
}
//到这里可以提取页面上的图片了
html = await globalData.Browser.GetHtmlSource();
var list = await HtmlAgilityPackUtil.GetImgFromHtmlAsync(html);
this.Dispatcher.Invoke(() => {
this.listbox_ImageDynamic.ItemsSource = list;
});
}
private async void LoadHotSpots()
{
//不加载
if (this.grid_Content.Children.Count > 0)
{
return;
}
List <TagImg> hotSpotsImgList = new List <TagImg>();
this.Dispatcher.BeginInvoke(new Action(() => {
dialog = new WaitingDailog("正在加载每日热图");
dialog.ShowDialog();
}));
hotSpotsImgList = await HtmlAgilityPackUtil.GetBingImgFromUrlAsync(UrlUtil.CNBingImageUrl, true);
//去除
hotSpotsImgList = hotSpotsImgList.Where(x => x.Src.Contains("tse1-mm")).ToList();
//显示
ShowImage(hotSpotsImgList, true);
dialog.Close();
}
