public static ArticleDocument GetArticle(string url, ref bool isGetContentSuccess)
{
try
{
Html html = Html.CreatHtml(new Uri(url));
if (html == null)
{
return(null);
}
ArticleDocument doc = GetRequestArticle(html.SourceCode);
isGetContentSuccess = doc.Content.Length > 0;
if (doc != null)
{
doc.Url = url;
doc.Content = doc.Content.Length > 0 ? doc.Content : html.Content;
doc.Timestamp = html.Timestamp;
doc.Site = GetSiteName(html.Title);
doc.ChildrenLink = html.ChildrenLink.Select(o => o.ToString()).ToList();
}
return(doc);
}
catch (Exception exception)
{
Console.WriteLine(exception.Message);
}
return(null);
}
protected void gvDocuments_RowDeleting(object sender, GridViewDeleteEventArgs e)
{
//Populate dto from session
GetSessionValues();
if (ArticleBF.ArticleDto.Documents[e.RowIndex].IsOld)
{
//Mark for deletion from database
ArticleBF.ArticleDto.Documents[e.RowIndex].IsDeleted = true;
}
else
{
//Get the document
ArticleDocument artDoc = ArticleBF.ArticleDto.Documents[e.RowIndex];
try
{
File.Delete(artDoc.FullFileName);
}
catch (IOException ioEx)
{
DisplayPopup("An error occured deleting the document.");
}
//Remove from documents list
ArticleBF.ArticleDto.Documents.RemoveAt(e.RowIndex);
}
//Update session
SetSessionValues();
//Reload the grid
LoadDocGrid();
}
public void ConvertHtmlToDoc(int facultyId, int semesterId)
{
var listArticleApproved =
_articleRepository.GetListArticlesApprovedAfterFinalDateByFaculty(facultyId, semesterId);
foreach (var article in listArticleApproved)
{
var articleDocInDb = _dbContext.ArticleDocuments.Where(d => article.Id.Equals(d.ArticleId)).ToList();
if (articleDocInDb.Any())
{
continue;
}
var fileName = Guid.NewGuid() + ".docx";
var docPath = GetDocPath(fileName);
ConvertDocument(article.Content, docPath);
var articleDoc = new ArticleDocument()
{
ArticleId = article.Id,
DocumentFile = fileName
};
_dbContext.ArticleDocuments.Add(articleDoc);
_dbContext.SaveChanges();
}
}
public void InitCommand(Database db, Int32 ArticleId, ArticleDocument artDoc)
{
dbCommand = db.GetStoredProcCommand(sqlCommand);
db.AddInParameter(dbCommand, "ArticleId", DbType.Int32, ArticleId);
db.AddInParameter(dbCommand, "ArtDocTypeId", DbType.Int32, artDoc.ArtDocTypeId);
db.AddInParameter(dbCommand, "DocId", DbType.Int32, artDoc.DocId);
}
public void InitCommand(Database db, Int32 articleId, ArticleDocument artDoc)
{
dbCommand = db.GetStoredProcCommand(sqlCommand);
db.AddInParameter(dbCommand, "ArticleId", DbType.Int32, articleId);
db.AddInParameter(dbCommand, "ArtDocTypeId", DbType.Int32, artDoc.ArtDocTypeId);
db.AddInParameter(dbCommand, "Data", DbType.Binary, artDoc.Data);
db.AddInParameter(dbCommand, "FileName", DbType.String, artDoc.FileName);
db.AddInParameter(dbCommand, "FileSizeKB", DbType.Int32, artDoc.CalculateFileSizeInKB());
db.AddInParameter(dbCommand, "Comments", DbType.String, artDoc.Comments);
db.AddInParameter(dbCommand, "Active", DbType.Int32, ConvBoolToInt32(artDoc.IsActive));
db.AddInParameter(dbCommand, "CreationUserId", DbType.Int32, artDoc.CreationUserId);
}
private ArticleDocument GetDocValues()
{
ArticleDocument artDoc = new ArticleDocument();
artDoc.ArtDocTypeId = Convert.ToInt32(ddlArtDocTypes.SelectedValue);
artDoc.Comments = tbComments.Text.Trim();
artDoc.CreationUserId = UserInfo.UserDto.UserId;
artDoc.IsActive = Convert.ToBoolean(Convert.ToInt32(ddlDocStatus.SelectedValue));
return(artDoc);
}
/// <summary>
/// 从给定的Html原始文本中获取正文信息
/// </summary>
/// <param name="html"></param>
/// <returns></returns>
private static ArticleDocument GetRequestArticle(string html)
{
// 如果换行符的数量小于10，则认为html为压缩后的html
// 由于处理算法是按照行进行处理，需要为html标签添加换行符，便于处理
if (html.Count(c => c == '\n') < 10)
{
html = html.Replace(">", ">\n");
}
// 获取html，body标签内容
string body = "";
string bodyFilter = @"(?is)<body.*?</body>";
Match m = Regex.Match(html, bodyFilter);
if (m.Success)
{
body = m.ToString();
}
// 过滤样式，脚本等不相干标签
foreach (var filter in Html2Article.Filters)
{
body = Regex.Replace(body, filter[0], filter[1]);
}
// 标签规整化处理，将标签属性格式化处理到同一行
// 处理形如以下的标签：
// <a
// href='http://www.baidu.com'
// class='test'
// 处理后为
// <a href='http://www.baidu.com' class='test'>
body = Regex.Replace(body, @"(<[^<>]+)\s*\n\s*", FormatTag);
string content;
string contentWithTags;
string title1;
GetContent(body, out content, out title1, out contentWithTags);
string title2 = GetTitle(html);
string title = GetMostProTitle(title1, title2, content);
ArticleDocument article = new ArticleDocument
{
Title = title,
PublishDate = GetPublishDate(html),
Content = content,
ContentWithTags = contentWithTags
};
return(article);
}
public async Task <ActionResult> Search(string s, int i = 0, int ps = 25)
{
SearchModel m = new SearchModel()
{
SettingsPath = Server.MapPath("~/articles/settings.json")
};
m.SearchText = s;
var watch = System.Diagnostics.Stopwatch.StartNew();
m.SearchResults = await ArticleDocument.GetArticleDocumentsAsync(s, i, ps);
watch.Stop();
m.SearchTime = watch.Elapsed.TotalSeconds;
return(View("~/Views/Articles/Search.cshtml", m));
}
public bool UploadDocument(IFormFileCollection files, int articleId)
{
var result = true;
if (files.Count == 0)
{
return(false);
}
foreach (var formFile in files)
{
var upFileName = formFile.FileName;
var fileName = Guid.NewGuid() + Path.GetExtension(upFileName);
var saveDir = Path.Combine(_webHostEnvironment.WebRootPath, "files");
var savePath = Path.Combine(saveDir, fileName);
try
{
if (!Directory.Exists(saveDir))
{
Directory.CreateDirectory(saveDir);
}
using var fs = File.Create(savePath);
formFile.CopyTo(fs);
fs.Flush();
var newDocument = new ArticleDocument
{
ArticleId = articleId,
DocumentFile = fileName
};
_dbContext.ArticleDocuments.Add(newDocument);
_dbContext.SaveChanges();
}
catch (Exception)
{
result = false;
}
}
return(result);
}
/// <summary>
/// Indexes an Article in Azure search
/// </summary>
/// <param name="article"></param>
/// <returns></returns>
static async Task IndexArticlesAsync(List <Article> articles)
{
List <ArticleDocument> list = new List <ArticleDocument>();
// create the index if it does not exist
Search.CreateIndex();
// Convert to ArticleDocument
foreach (Article a in articles)
{
if (null == a)
{
continue;
}
ArticleDocument d = new ArticleDocument();
d.FromArticle(a);
list.Add(d);
}
await Search.MergeOrUploadAsync(list);
}
protected void btnAddDoc_Click(object sender, EventArgs e)
{
if (Page.IsValid)
{
//Get session values
GetSessionValues();
try
{
if (tbUploadDocument.HasFile && tbUploadDocument.PostedFile != null)
{
string fileName = SaveDocToTemp(tbUploadDocument);
ArticleDocument artDoc = GetDocValues();
artDoc.FullFileName = fileName;
ArticleBF.ArticleDto.Documents.Add(artDoc);
SetSessionValues();
LoadDocGrid();
ResetDocumentFormValues();
}
else
{
DisplayPopup("The document cannot be empty. Please choose a document with content.");
}
}
catch (IOException ioEx)
{
throw;
}
}
}
protected void gvDocuments_RowDataBound(object sender, GridViewRowEventArgs e)
{
if (e.Row.RowType.Equals(DataControlRowType.DataRow))
{
if (ArticleBF.ArticleDto.Documents[e.Row.RowIndex].IsDeleted)
{
e.Row.Visible = false;
}
Label lblDocType = (Label)e.Row.FindControl("lblDocType");
ArticleDocument artDoc = (ArticleDocument)e.Row.DataItem;
if (!(lblDocType == null || artDoc == null))
{
ListItem liItem = ddlArtDocTypes.Items.FindByValue(artDoc.ArtDocTypeId.ToString());
if (liItem != null)
{
lblDocType.Text = liItem.Text;
}
}
}
}
public static Article AdminGetArticle(Int32 ArticleId)
{
Article art = new Article();
try
{
DataSet articleSet = ArticleQueryHelper.AdminGetArticle(ArticleId);
if (articleSet.Tables.Count == 2)
{
//Get the issue details
if (articleSet.Tables[0].Rows.Count > 0)
{
DataRow artRow = articleSet.Tables[0].Rows[0];
art.ArticleId = (Int32)artRow["ArticleId"];
art.Title = (string)artRow["Title"];
art.Authors = (string)artRow["Authors"];
art.Keywords = (string)artRow["Keywords"];
art.PageNumber = (Int32)artRow["PageNumber"];
art.IsActive = (bool)artRow["Active"];
}
//Get the documents
if (articleSet.Tables[1].Rows.Count > 0)
{
ArticleDocument artDoc;
art.Documents = new List <ArticleDocument>();
foreach (DataRow docRow in articleSet.Tables[1].Rows)
{
artDoc = new ArticleDocument();
artDoc.DocId = (Int32)docRow["DocId"];
artDoc.FullFileName = (string)docRow["FileName"];
artDoc.ArtDocTypeDescription = (string)docRow["ArtDocTypeDescription"];
artDoc.ArtDocTypeId = (Int32)docRow["ArtDocTypeId"];
artDoc.IsActive = (bool)docRow["Active"];
artDoc.IsNew = false;
art.Documents.Add(artDoc);
}
}
}
return(art);
}
catch (SqlException sqlEx)
{
Console.WriteLine(sqlEx.ToString());
throw new DataException("An exception occured getting the article details.", sqlEx);
}
}
