/// <summary>
/// get detail article
/// </summary>
/// <param name="articleUrl"></param>
/// <returns></returns>
public RemoteArticleInfo GetRemoteArticleInfo(string articleUrl)
{
SiteDetect(articleUrl);
// get article htmlnode
ArticleElementInfo articleElement = Site?.GetArticleElement(articleUrl);
if (articleElement == null)
{
return(null);
}
if (articleElement.Content != null)
{
articleElement.Content = Utilities._ClearHtmlTag(articleElement.Content);
}
var remoteArticleInfo = new RemoteArticleInfo();
//set content
//domain
Uri uri = new Uri(articleUrl);
remoteArticleInfo.domain = uri.Host;
//url
remoteArticleInfo.url = articleUrl;
//title
remoteArticleInfo.title = articleElement.Title != null?HttpUtility.HtmlDecode(articleElement.Title.InnerText.Trim()) : string.Empty;
//_excerpt
if (articleElement.Excerpt != null)
{
remoteArticleInfo.excerpt = string.IsNullOrEmpty(HttpUtility.HtmlDecode(articleElement.Excerpt.InnerText.Trim())) ? articleElement.Excerpt.Attributes["content"].Value.Trim() : HttpUtility.HtmlDecode(articleElement.Excerpt.InnerText.Trim());
}
else
{
remoteArticleInfo.excerpt = string.Empty;
}
//content
remoteArticleInfo.content = articleElement.Content != null?HttpUtility.HtmlDecode(articleElement.Content.InnerHtml) : string.Empty;
//author
remoteArticleInfo.author = articleElement.Author != null?HttpUtility.HtmlDecode(articleElement.Author.InnerText.Trim()) : string.Empty;
//_image
if (articleElement.Image != null)
{
remoteArticleInfo.lead_image_url = articleElement.Image.Attributes["src"]?.Value.Trim() ?? articleElement.Image.Attributes["content"].Value.Trim();
if (remoteArticleInfo.lead_image_url == null)
{
return(null);
}
}
//publish time
remoteArticleInfo.date_published = articleElement.PublishedTime != null?HttpUtility.HtmlDecode(articleElement.PublishedTime.InnerText.Trim()) : string.Empty;
//keyword
if (articleElement.Keyword != null)
{
for (int i = 0; i < articleElement.Keyword.Count &&
(remoteArticleInfo.keyword + articleElement.Keyword[i].InnerText.Trim().Replace("#", "")).Length < 255; i++)
{
string xxx = HttpUtility.HtmlDecode(articleElement.Keyword[i].InnerText.Trim().Replace("#", "").Replace("&2", ""));
remoteArticleInfo.keyword += xxx;
if (i != articleElement.Keyword.Count - 1)
{
remoteArticleInfo.keyword += ',';
}
}
}
else
{
remoteArticleInfo.keyword = string.Empty;
}
//replace
remoteArticleInfo.content = Utilities._ClearContent(remoteArticleInfo.content);
var replaceObj = Site.GetStringToReplace();
if (replaceObj != null)
{
foreach (var obj in replaceObj)
{
remoteArticleInfo.content = remoteArticleInfo.content.Replace(obj.ToString(), string.Empty);
}
}
return(remoteArticleInfo);
}
static void InsertArticle(string username, RemoteArticleInfo article, string category)
{
if (article.url.Contains("http://bestie.vn/") || article.url.Contains("http://thegioitre.vn/"))
{
article.content = System.Net.WebUtility.HtmlDecode(article.content);
article.keyword = string.Empty;
article.author = article.author.Split('-')[0].Trim();
}
if (article.author.Length > 64)
{
article.author = string.Empty;
}
if (string.IsNullOrEmpty(article.keyword))
{
article.keyword = "";
}
if (article.keyword.Length > 255)
{
string tmp = "";
for (int i = 0; i < article.keyword.Split(',').Length; i++)
{
if ((tmp + article.keyword.Split(',')[i]).Length < 255)
{
tmp += article.keyword.Split(',')[i];
}
}
article.keyword = tmp;
}
if (string.IsNullOrEmpty(article.lead_image_url) ||
string.IsNullOrEmpty(article.content) ||
string.IsNullOrEmpty(article.title) ||
string.IsNullOrWhiteSpace(article.title) ||
string.IsNullOrWhiteSpace(article.content)
)
{
return;
}
if (article.excerpt.Length > 999)
{
article.excerpt = article.excerpt.Substring(0, 999);
}
//headline re-validate
article.excerpt = RevalidateHeadline(article.excerpt);
if (string.IsNullOrEmpty(article.keyword))
{
article.keyword = string.Empty;
}
article.date_published = article.date_published.Replace(" ", " ");
article.date_published = DateTimeParse(article.date_published.Trim());
//publish time
article.date_published = DateTime.TryParse(article.date_published.Trim(), out var publishTime) ? publishTime.ToString(CultureInfo.InvariantCulture) : DateTime.Now.ToString(CultureInfo.InvariantCulture);
//author
if (!string.IsNullOrEmpty(article.author))
{
article.author = RemoveAtEnd(article.author);
}
string frendlyTitle = VietCMS.Framework.Core.Common.WebControl.ToFriendlyString(article.title);
if (_appDbContext.Articles.Count(x => x.SourceUrl.Equals(article.url) ||
x.FriendlyTitle.Equals(frendlyTitle)) == 0)
{
string tmp = "";
if (string.IsNullOrEmpty(article.keyword))
{
article.keyword = String.Empty;
}
string[] s = article.keyword.Split(',');
List <Keyword> keywords = new List <Keyword>();
for (int i = 0; i < s.Length; i++)
{
string si = s[i];
if (si.Trim().Length > 40 || si.Trim().Length < 2)
{
continue;
}
tmp += VietCMS.Framework.Core.Common.WebControl.ToFriendlyString(si);
if (i != s.Length - 1)
{
tmp += ",";
}
if (_appDbContext.Keywords.Count(x => x.Title.Equals(si.Trim())) == 0)
{
keywords.Add(_appDbContext.Keywords.Add(new Keyword()
{
Title = s[i],
CreatedAt = DateTime.Now,
FriendlyTitle = VietCMS.Framework.Core.Common.WebControl.ToFriendlyString(si),
Type = "Keyword",
VisitCount = 0
}));
}
else
{
keywords.Add(_appDbContext.Keywords.FirstOrDefault(x => x.Title.Equals(si.Trim())));
}
}
_appDbContext.SaveChanges();
List <RelatedArticle> relatedArticles = new List <RelatedArticle> ();
foreach (Keyword k in keywords)
{
List <Article> articles = _appDbContext.ArticleKeywords
.Where(y => y.Keyword.Id == k.Id)
.Take(20)
.OrderByDescending(x => x.Article.CreatedAt)
.Select(x => x.Article)
.ToList();
foreach (Article item in articles)
{
bool check = false;
foreach (RelatedArticle ra in relatedArticles)
{
if (item.Id == ra.Origin.Id)
{
ra.Index++;
check = true;
}
}
if (!check)
{
RelatedArticle relatedArticle = new RelatedArticle()
{
Origin = item,
Index = 1,
CreatedAt = DateTime.Now,
Type = "Related"
};
relatedArticles.Add(relatedArticle);
}
}
}
Article a = new Article()
{
Title = article.title.Trim(),
Thumbnail = article.lead_image_url,
Category = _appDbContext.Categories.FirstOrDefault(x => x.FriendlyName.Equals(category)),
Headlines = article.excerpt.Trim(),
Content = article.content,
Keywords = article.keyword.Trim(),
FriendlyKeywords = tmp,
Source = article.domain,
SourceUrl = article.url,
FriendlyTitle = VietCMS.Framework.Core.Common.WebControl.ToFriendlyString(article.title.Trim()),
AuthorAlias = article.author.Trim(),
Status = Const.ArticleStatusActive,
CreatedAt = DateTime.Now,
CreatedBy = _appDbContext.Users.FirstOrDefault(x => x.Username.Equals(username)),
LastModifiedBy = _appDbContext.Users.FirstOrDefault(x => x.Username.Equals(username)),
LastModifiedAt = DateTime.Now,
ViewCount = 0,
SubTitle = article.title.Trim().Length > 50 ? article.title.Trim().Substring(0, 47) + "..." : article.title.Trim()
};
foreach (Keyword keyword in keywords)
{
_appDbContext.ArticleKeywords.Add(new ArticleKeyword()
{
Article = a,
Keyword = keyword
});
}
for (int i = 0; i < relatedArticles.Count; i++)
{
var r = relatedArticles[i];
r.Related = a;
_appDbContext.RelatedArticles.Add(r);
_appDbContext.RelatedArticles.Add(new RelatedArticle()
{
CreatedAt = DateTime.Now,
Index = r.Index,
Origin = a,
Related = r.Origin,
Type = "Related"
});
}
_appDbContext.Articles.Add(a);
AddedArticles.Add(a);
_appDbContext.SaveChanges();
}
}
