/// <summary>
/// 初始化。
/// </summary>
public ItemChapter(Models.Domains.Entities.Chapter target)
{
this.ChapterId = target.ChapterId;
this.ChapterName = target.ChapterName;
this.Priority = target.Priority;
}
/// <summary>
/// 获取域对象。
/// </summary>
/// <returns>域对象。</returns>
public Models.Domains.Entities.Chapter GetReturn()
{
var target = new Models.Domains.Entities.Chapter();
target.ChapterId = Guid.NewGuid();
target.DocumentId = this.DocumentId;
target.ChapterName = this.ChapterName.Trim();
target.Priority = this.Priority;
target.UpdateTime = DateTime.Now;
return(target);
}
public override bool ParsePage()
{
this.Document = new Models.Domains.Entities.Document();
var DOM = new HtmlAgilityPack.HtmlDocument();
DOM.LoadHtml(this.SourceText);
try
{
var div = DOM.GetElementbyId("main");
var h2s = div.MyChildNodeFilter("h2");
var i = 1;
foreach (var h2 in h2s)
{
if (h2.Attributes["class"].Value == "hiddenAnchor")
{
continue;
}
var name = h2.InnerText;
if (string.IsNullOrEmpty(name))
{
continue;
}
var chapter = new Models.Domains.Entities.Chapter();
chapter.ChapterId = Guid.NewGuid();
chapter.DocumentId = this.DocumentId;
chapter.ChapterName = name;
chapter.Priority = i++;
chapter.UpdateTime = DateTime.Now;
this.Document.Chapters.Add(chapter);
}
}
catch
{
return(false);
}
return(true);
}
public List <Models.Domains.Entities.Chapter> GetReturnList()
{
var db = new Models.Domains.Entities.DMsDbContext();
var document = db.Documents.Find(this.DocumentId);
int maxPriority;
if (document.Chapters.Count() == 0)
{
maxPriority = 0;
}
else
{
maxPriority = document.Chapters.Max(c => c.Priority);
}
var listInput = new List <string>(this.Input.Split(new string[] { "\n" }, StringSplitOptions.RemoveEmptyEntries));
var listChapter = new List <Models.Domains.Entities.Chapter>();
foreach (var itemInput in listInput)
{
if (string.IsNullOrEmpty(itemInput))
{
continue;
}
var itemChapter = new Models.Domains.Entities.Chapter();
itemChapter.ChapterId = Guid.NewGuid();
itemChapter.DocumentId = this.DocumentId;
itemChapter.ChapterName = itemInput.Trim();
itemChapter.Priority = ++maxPriority;
itemChapter.UpdateTime = DateTime.Now;
listChapter.Add(itemChapter);
}
return(listChapter);
}
public override bool ParsePage()
{
this.Document = new Models.Domains.Entities.Document();
var DOM = new HtmlAgilityPack.HtmlDocument();
DOM.LoadHtml(this.SourceText);
try
{
//var meta = DOM.DocumentNode.MyGetNode("html",1).MyGetNode("head", 1).MyGetNode("meta", "name", "APIName", 1);
var div_main = DOM.GetElementbyId("main");
var div_main_h1 = div_main.MyGetNode("h1", 1);
this.Document.DocumentId = Guid.NewGuid();
//this.Document.ParentDocumentId
this.Document.Title = div_main_h1.InnerText;
this.Document.Title = System.Web.HttpUtility.HtmlDecode(this.Document.Title);
this.Document.Title = System.Text.RegularExpressions.Regex.Replace(this.Document.Title, @"\s+", " ").Trim();
//this.Document.NodeName = meta.Attributes["content"].Value;
//this.Document.NodeName = System.Web.HttpUtility.HtmlDecode(this.Document.NodeName);
{
switch (this.NodeNameParsingMode)
{
case "MicrosoftDocsApiType":
this.Document.NodeName = this.Document.Title.Split(new string[] { " " }, StringSplitOptions.None)[0];
break;
case "MicrosoftDocsApiMember":
this.Document.NodeName = this.Document.Title.Split(new string[] { " ", "." }, StringSplitOptions.None)[1];
break;
default:
break;
}
}
//this.Document.Priority
this.Document.UpdateTimeForHTTPGet = DateTime.Now;
this.Document.Url = this.Url;
this.Document.UpdateTime = DateTime.Now;
//处理Chapters
var h2s = div_main.Descendants("h2");
var i = 1;
foreach (var h2 in h2s)
{
if (h2.Attributes["class"].Value == "hiddenAnchor")
{
continue;
}
var name = h2.InnerText;
if (string.IsNullOrEmpty(name))
{
continue;
}
var chapter = new Models.Domains.Entities.Chapter();
chapter.ChapterId = Guid.NewGuid();
chapter.DocumentId = this.Document.DocumentId;
chapter.ChapterName = name;
chapter.Priority = i++;
chapter.UpdateTime = DateTime.Now;
this.Document.Chapters.Add(chapter);
}
this.Document.IsGetAllChapters = true;
}
catch
{
return(false);
}
return(true);
}
