public static async Task LoadItemsFromTag(NewsItemList plist, ItemTag tag)
{
if (tag == null)
{
return;
}
await DoSearch(plist, tag.Title);
}
public static async Task DoSearch(NewsItemList plist, string keyword)
{
try
{
plist.Clear();
string url = "http://timkiem.vnexpress.net/?q=" + keyword;
string html = await HtmlDownloader.loadFromUrl(url);
HtmlDocument page = new HtmlDocument();
page.LoadHtml(html);
var allResultNodes = page.DocumentNode.SelectNodes("//li[@class='block_search_result_text']");
if (allResultNodes == null)
{
MessageBox.Show("Sorry, no articles found!");
return;
}
foreach (HtmlNode itemNode in allResultNodes)
{
NewsItem article = new NewsItem();
article.Source = "Vnexpress";
HtmlNode nodeLink = itemNode.SelectSingleNode(itemNode.XPath + "//a[@href]");
HtmlNode nodeImage = itemNode.SelectSingleNode(itemNode.XPath + "//img[@src]");
if (nodeImage != null)
{
article.ImageLink = nodeImage.Attributes["src"].Value;
}
article.Name = HtmlDownloader.removeHtml(nodeLink.Attributes["alt"].Value);
article.LinkUrl = nodeLink.Attributes["href"].Value;
HtmlNode nodeTime = itemNode.SelectSingleNode(itemNode.XPath + "//p[@class='txt_gray txt_11 ex_hi']");
article.DatePublished = HtmlDownloader.removeHtml(nodeTime.InnerHtml);
HtmlNode nodeShortContent = itemNode.SelectSingleNode(itemNode.XPath + "//span[@class='hightlight']");
article.ShortContent = HtmlDownloader.removeHtml(nodeShortContent.InnerText);
plist.Add(article);
}
if (plist.Count == 0)
{
MessageBox.Show("Sorry, no articles found!");
}
}
catch (Exception ex)
{
MessageBox.Show("System got an error at DoSearch function with message:\n" + ex.Message);
return;
}
}
public static async Task DoSearch(NewsItemList plist, string keyword)
{
try
{
plist.Clear();
string url = "http://search.dantri.com.vn/SearchResult.aspx?s=" + keyword + "&PageIndex=1";
await LoadItemsFromPage(plist, url, "Dantri", null);
if (plist.Count == 0)
{
MessageBox.Show("Sorry, no articles found!");
}
}
catch (Exception ex)
{
MessageBox.Show("System got an error at DoSearch function with message:\n" + ex.Message);
return;
}
}
//load articles from source, category, url
public static async Task LoadItemsFromPage(NewsItemList plist, string url, string source, NewsItem hottest = null)
{
try
{
if (!url.Contains("vnexpress.net"))
{
url = "http://vnexpress.net" + url;
}
plist.Clear();
if (hottest != null)
{
hottest.Source = source;
}
string html = await HtmlDownloader.loadFromUrl(url);
HtmlDocument page = new HtmlDocument();
page.LoadHtml(html);
var allItemNodes = page.DocumentNode.SelectNodes("//div[@class='folder-news']");
getHottestNew(page, hottest);
if (hottest != null)
{
plist.Add(hottest);
}
foreach (HtmlNode itemNode in allItemNodes)
{
NewsItem article = new NewsItem();
article.Source = source;
HtmlNode nodeImage = itemNode.SelectSingleNode(itemNode.XPath + "//img[@src]");
if (nodeImage != null)
{
article.ImageLink = nodeImage.Attributes["src"].Value;
}
HtmlNode nodeTitle = itemNode.SelectSingleNode(itemNode.XPath + "//a[@class='link-title14' and @href]");
article.Name = HtmlDownloader.removeHtml(nodeTitle.InnerText);
article.LinkUrl = nodeTitle.Attributes["href"].Value;
HtmlNode nodeTime = itemNode.SelectSingleNode(itemNode.XPath + "//span[@class='timeListHome']");
if (nodeTime != null)
{
article.DatePublished = HtmlDownloader.removeHtml(nodeTime.InnerText);
}
HtmlNode nodeShortContent = itemNode.SelectSingleNode(itemNode.XPath + "//h3[@class='h3Lead']");
if (nodeShortContent != null)
{
//cut the related link out
int cutIndex = nodeShortContent.InnerHtml.IndexOf("<br");
string shortContent = nodeShortContent.InnerHtml;
//if find the trash
if (cutIndex > 0)
{
shortContent = nodeShortContent.InnerHtml.Substring(0, cutIndex);
}
//modify the trash in the content
shortContent = shortContent.Replace(">", "");
article.ShortContent = HtmlDownloader.removeHtml(shortContent);
}
plist.Add(article);
}
}
catch (Exception ex)
{
MessageBox.Show("System got an error at LoadItemsFromPage function with message:\n" + ex.Message);
return;
}
}
//hàm lấy một list các article từ 1 url của category
public static async Task LoadItemsFromPage(NewsItemList plist, string url, string source, NewsItem hottest = null)
{
try
{
if (!url.Contains("http://"))
{
url = "http://dantri.com.vn" + url;
}
plist.Clear();
if (hottest != null)
{
hottest.Source = source;
}
string html = await HtmlDownloader.loadFromUrl(url);
HtmlDocument page = new HtmlDocument();
page.LoadHtml(html);
var allItemNodes = page.DocumentNode.SelectNodes("//div[@class='mt3 clearfix']");
if (getHottestNew(page, hottest))
{
if (hottest != null)
{
plist.Add(hottest);
}
}
if (allItemNodes == null)
{
return;
}
foreach (HtmlNode itemNode in allItemNodes)
{
NewsItem article = new NewsItem();
article.Source = source;
HtmlNode nodeImage = itemNode.SelectSingleNode(itemNode.XPath + "//img[@src]");
if (nodeImage != null)
{
article.ImageLink = nodeImage.Attributes["src"].Value;
}
var nodeTitle = itemNode.SelectNodes(itemNode.XPath + "//a");
foreach (HtmlNode node in nodeTitle)
{
if (node.Attributes.Contains("href"))
{
if (article.LinkUrl == null || article.LinkUrl == "")
{
article.LinkUrl = node.Attributes["href"].Value;
}
}
if (node.Attributes.Contains("title"))
{
if (article.Name == null || article.Name == "")
{
article.Name = HtmlDownloader.removeHtml(node.Attributes["title"].Value);
}
}
if (!node.InnerHtml.Contains("<img"))
{
if (node.InnerText.Length > 10)
{
if (article.Name == null || article.Name == "")
{
article.Name = HtmlDownloader.removeHtml(node.InnerText);
}
}
}
}
if (article.LinkUrl.Contains("tuyensinh.dantri"))
{
continue;
}
HtmlNode nodeShortContent = itemNode.SelectSingleNode(itemNode.XPath + "//div[@class='fon5 wid324 fl']");
if (nodeShortContent == null)
{
nodeShortContent = itemNode.SelectSingleNode(itemNode.XPath + "//div[@class='fon5 fl']");
}
if (nodeShortContent != null)
{
//cut the related link out
int cutIndex = nodeShortContent.InnerHtml.IndexOf("<br");
string shortContent = nodeShortContent.InnerHtml;
//if find the trash
if (cutIndex > 0)
{
shortContent = nodeShortContent.InnerHtml.Substring(0, cutIndex);
}
//modify the trash in the content
shortContent = shortContent.Replace(">", "");
article.ShortContent = HtmlDownloader.removeHtml(shortContent);
}
plist.Add(article);
}
}
catch (Exception ex)
{
MessageBox.Show("System got an error at LoadItemsFromPage function with message:\n" + ex.Message);
return;
}
}
public static async Task LoadItemsFromTag(NewsItemList plist, ItemTag tag)
{
await LoadItemsFromPage(plist, tag.Link, "Dantri", null);
}
