private async Task <New> GetContents(LinkObj obj)
{
var doc = await GetDocuments(obj.Link);
var title = doc.DocumentNode.SelectSingleNode("/html/body/section[4]/div[2]/h1");
var contents = doc.DocumentNode.SelectSingleNode("/html/body/section[4]/div[2]/article");
var auth = doc.DocumentNode.SelectSingleNode("/html/body/section[4]/div[2]/div[1]/div[1]/div/strong");
return(new New(title.InnerText, contents.InnerHtml, "", "ndh.vn" + "-" + Type + "-" + auth.InnerText, obj.Img, this.Categories));
}
private async Task <New> GetContents(LinkObj obj)
{
var doc = await GetDocuments(obj.Link);
var title = doc.DocumentNode.QuerySelector("body > div.wrap-main > div.container.wrap-main-page > div.left-col > div.block.pb15 > div.box-post-main-title.pb15 > h1");
var contents = doc.DocumentNode.QuerySelector("#sevenBoxNewContentInfo");
var auth = doc.DocumentNode.QuerySelector("body > div.wrap-main > div.container.wrap-main-page > div.left-col > div:nth-child(2) > div.page-content > div.sevenPostWrap.pb10 > div.sevenPostAuthor");
var img = GetImage(contents.InnerHtml);
return(new New(title.InnerText, contents.InnerHtml, "", "cafeland.vn" + "-" + Type + "-" + auth.InnerText, img, this.Categories));
}
private async Task <List <LinkObj> > GetLinks()
{
HtmlDocument docs = await GetDocuments(this.Url);
var childArticles = docs.DocumentNode.SelectNodes("//*[@id=\"dnn_ctr571_ModuleContent\"]/div/div[1]/div/div[1]/a");
// var childArticles = docs.DocumentNode.SelectNodes("/html/body/div[1]/div/div[4]/div[1]/section/div/article/figure/a") == null ?
// docs.DocumentNode.SelectNodes("/html/body/div[1]/div/div[3]/div[1]/div/section/div/article/figure/a") : null;
List <LinkObj> links = new List <LinkObj>();
// links.Add(mainArticle);
foreach (var a in childArticles)
{
var link = new LinkObj();
link.Link = a.Attributes["href"].Value;
var img = a.ChildNodes.FirstOrDefault(e => e.Name == "img");
if (img != null)
{
link.Img = "http://tiasang.com.vn" + img.Attributes["src"].Value;
}
links.Add(link);
}
return(links);
}
private async Task <New> GetContents(LinkObj link)
{
var doc = await GetDocuments(link.Link);
string title = "", description = "", author = "", source = string.Empty;
title = doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/h1") != null?doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/h1").InnerText : string.Empty;
description = doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/p") != null?doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/p").InnerText : string.Empty;
author = doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/article/p/strong") != null?doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/article/p/strong").InnerText : string.Empty;
source = doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/article/p/em") != null?doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/article/p/em").InnerText : string.Empty;
var rendered = doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]");
var desc = rendered.ChildNodes[5];
var realDesc = "";
if (desc != null)
{
foreach (var item in desc.ChildNodes)
{
realDesc += item.InnerText + "-";
}
realDesc = realDesc.Remove(realDesc.Length - 1);
}
var arti = rendered.ChildNodes[7];
var contents = "";
var artChilds = arti.ChildNodes.Where(e => e.Name == "p" || e.Name == "table").ToList();
for (int i = 0; i < artChilds.Count; i++)
{
if (artChilds[i].InnerHtml.Contains("<strong>"))
{
continue;
}
contents += artChilds[i].OuterHtml;
}
var auth = arti.ChildNodes[arti.ChildNodes.Count - 2];
var img = doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/article/table/tbody/tr[1]/td/img");
string url = img == null ? link.Img : img.Attributes["src"].Value;
url = url.Split("_")[0];
if (!url.Contains(".jpg"))
{
url += ".jpg";
}
arti.RemoveChild(auth);
var realAuth = "";
if (auth.Name == "p")
{
realAuth = auth.InnerText.Trim();
}
else
{
realAuth = author.Trim();
}
if (realAuth != "")
{
realAuth += "-" + source.Trim();
}
else
{
realAuth += source;
}
// var auth = doc.DocumentNode.SelectSingleNode("/html/body/section[2]/section[1]/section[1]/article/p[contains(@style,'text-align:right;')]");
// rendered.RemoveChild(auth);
var newRender = realDesc + contents;
return(new New
(
title,
this.RemoveLink(newRender), "",
"vnexpress.net-" + realAuth,
url,
this.Categories
));
}
