public async Task <List <ImageModel> > FindImageUrls(ImagePageModel imagePageModel)
{
string html = await HttpDownloader.DownloadHtmlPage(imagePageModel.ImagePageUrl);
IBrowsingContext context = BrowsingContext.New();
IDocument document = await context.OpenAsync(response => response.Content(html));
List <ImageModel> imageModels = new List <ImageModel>();
IElement titleElement = document.QuerySelector("title");
string albumTitle = titleElement?.InnerHtml;
IElement divElement = document.QuerySelector("div.picture");
var imageElements = divElement.QuerySelectorAll("img");
if (imageElements == null || imageElements.Count() == 0)
{
Debug.WriteLine("在图片页面hmtl中没有匹配到图片url元素: " + imagePageModel.ImagePageUrl);
throw new Exception("在图片页面hmtl中没有匹配到图片url元素");
}
string pageNum = "";
IElement pageNumElement = document.QuerySelector("div.paging");
pageNum = pageNumElement?.QuerySelector("b")?.InnerHtml;
int index = 0;
foreach (IElement imgElement in imageElements)
{
string imageName;
string url = imgElement.GetAttribute("src");
if (index > 0)
{
imageName = albumTitle + "_" + pageNum + "_" + index;
}
else
{
imageName = albumTitle + "_" + pageNum;
}
ImageModel imageModel = new ImageModel(imagePageModel, imageName, url);
index++;
Debug.WriteLine("成功:找到图片" + imageModel.ImageUrl + imageModel.ImageName + imageModel.ImagePage.ImagePageUrl);
imageModels.Add(imageModel);
}
return(imageModels);
}
public async Task <List <ImageModel> > FindImageUrls(ImagePageModel imagePageModel)
{
string html = await HttpDownloader.DownloadHtmlPage(imagePageModel.ImagePageUrl);
IConfiguration config = Configuration.Default;
IBrowsingContext context = BrowsingContext.New(config);
IDocument document = await context.OpenAsync(response => response.Content(html));
IEnumerable <IElement> urlResult = document.All.Where(m => m.LocalName == "img" && m.ClassName == "lazy no_image");
string imageUrl;
if (urlResult.Count() == 0)
{
urlResult = document.All.Where(m => m.LocalName == "div" && m.ClassName == "image_1");
if (urlResult.Count() == 0)
{
Debug.WriteLine("失败：" + "解析失败，没有找到图片链接。");
throw new Exception("失败：" + "解析失败，没有找到图片链接。");
}
else
{
imageUrl = "https://" + urlResult.First().QuerySelector("img").GetAttribute("src").Remove(0, 2);
}
}
else
{
imageUrl = "https://" + urlResult.First().GetAttribute("src").Remove(0, 2);
}
string imageName = document.All.Where(m => m.LocalName == "title").First().Text();
ImageModel imageModel = new ImageModel(imagePageModel, imageName, imageUrl);
return(new List <ImageModel>()
{
imageModel
});
}
public async Task <List <ImageModel> > FindImageUrls(ImagePageModel imagePageModel)
{
string html = await HttpDownloader.DownloadHtmlPage(imagePageModel.ImagePageUrl);
IConfiguration config = Configuration.Default;
IBrowsingContext context = BrowsingContext.New(config);
IDocument document = await context.OpenAsync(response => response.Content(html));
var urlResult = document.All.Where(m => m.LocalName == "img" && m.Id == "img");
string imageUrl = urlResult.First().GetAttribute("src");
var imageNumResult = document.All.Where(m => m.LocalName == "div" && m.ClassName == "sn");
string imageNum = imageNumResult.First().QuerySelector("div").QuerySelector("span").Text();
var titleResult = document.All.Where(m => m.LocalName == "div" && m.Id == "i1");
string title = titleResult.First().QuerySelector("h1").Text();
string imageName = title + "_" + imageNum;
return(new List <ImageModel>()
{
new ImageModel(imagePageModel, imageName, imageUrl)
});
}
