/// <summary>
/// 生成目录
/// </summary>
/// <param name="book"></param>
public string UploadIndexFile(TygModel.书名表 book)
{
string dirserver = string.Format(createDirUrl, System.IO.Path.GetDirectoryName("/" + book.GetHTMLFilePath()));
//创建目录
string dic = GetWebDic(dirserver);
//生成目录成功
if (!string.IsNullOrEmpty(dic))
{
//生成没有成功
//移动当前工作目录
ftp.GotoDirectory(BaseFtpDic + Book.GetHTMLFilePath().Replace("/index.html", string.Empty), true);
string dirindex = BaseWebSite + "HTML/" + Book.分类标识.Trim().ToPingYing() + "/" + Book.书名.Trim().ToPingYing() + "/index.html";
//ftp上传文件
if (Upload(dirindex))
{
System.Console.WriteLine(string.Format("FTP {0},生成目录---------完成", Book.书名));
}
}
return(dic);
}
/// <summary>
/// 开始运行数据
/// </summary>
public void Run(Action callBack, TygModel.书名表 book)
{
//数据无效直接返回
if (book == null)
{
return;
}
//记时器
System.Diagnostics.Stopwatch watch = new System.Diagnostics.Stopwatch();
System.Threading.Tasks.Task task = System.Threading.Tasks.Task.Factory.StartNew(new Action(delegate
{
using (TygModel.Entities entity = new TygModel.Entities())
{
var itembook = entity.书名表.Where(b => b.GUID == book.GUID).FirstOrDefault();
CreateBookHTML createbook = new CreateBookHTML(itembook);
createbook.Run();
}
//线程写文件时不锁会冲突 另一个文件正在使用的异常
lock (lockobj)
{
//采集完成一个记录一个
System.IO.File.AppendAllLines(ProgressFileName, new string[] { book.GUID.ToString() + ((char)2) + book.书名 });
}
}));
try
{
task.Wait();
task.Dispose();
}
catch (Exception ex)
{
throw ex;
}
//进度加1
System.Threading.Interlocked.Increment(ref CurrentIndex);
//输出当前进度
System.Console.WriteLine(string.Format("生成完成{0}，用时:{1},进度:{2}/{3}", book.书名, Math.Round(watch.ElapsedMilliseconds / 1000.0 / 60, 2), CurrentIndex, books.Count));
watch.Stop();
//运行下一个数据
callBack();
}
/// <summary>
/// 得到文章的内容
/// </summary>
/// <param name="o">记录</param>
/// <param name="entities">数据库操作实体</param>
private void GetContents(TygModel.书名表 o, TygModel.Entities entities)
{
//得到
Skybot.Collections.Analyse.SingleListPageAnalyse ListPageAnalyse = null;
Skybot.Collections.Analyse.XMLDocuentAnalyse documentAnalyse = null;
//打印状态
System.Diagnostics.Debug.WriteLine("开始采集:" + o.书名 + " 目录:" + o.采集用的URL1);
//得到转换后的对应分析器
List <Task> tasks = new List <Task>();
tasks.Add(Task.Factory.StartNew(() =>
{
ListPageAnalyse = new Skybot.Collections.Analyse.SingleListPageAnalyse(o.采集用的URL1);
documentAnalyse = new Skybot.Collections.Analyse.XMLDocuentAnalyse()
{
IndexPageUrl = new Skybot.Collections.Analyse.ListPageContentUrl()
{
index = 0, Title = o.书名, Url = new Uri(o.采集用的URL1)
}
};
try
{
//初始化内容分析器
documentAnalyse.GetPathExpression(documentAnalyse.IndexPageUrl, ListPageAnalyse.ListPageContentUrls);
}
catch (Exception)
{
}
}));
try
{
//等待添加页面列表分析完成 5 分钟没有完成则表示超时
System.Threading.Tasks.Task.WaitAll(tasks.ToArray(), TimeSpan.FromMinutes(5));
}
catch (AggregateException ex)
{
ex.Handle((exx) =>
{
System.Diagnostics.Debug.WriteLine(exx.Message + "|||||" + exx.StackTrace);
return(true);
});
}
//如果数据有效
if (ListPageAnalyse != null && documentAnalyse != null && documentAnalyse.PathExpression != null)
{
//得到已经存在的记录
var docentitys = o.文章表.ToList();
var docs = docentitys.Select(p => p.章节名.Trim());
//打印状态
System.Diagnostics.Debug.WriteLine("内容共:" + ListPageAnalyse.ListPageContentUrls.Count + "条记录");
//将章节列表索引转换成为扩展实体数据
List <UrlExtentEntity> entitys = ListPageAnalyse.ListPageContentUrls.Select(p => new UrlExtentEntity()
{
index = p.index, Indexs = p.Indexs, Title = p.Title, Url = p.Url
}).ToList();
//开始采集数据
for (int k = 0; k < entitys.Count; k++)
{
//上一条记录
UrlExtentEntity PreviousItem = null; // entitys[k - 1];
//当前记录
UrlExtentEntity CurrentItem = null; // entitys[k];
//下一条记录
UrlExtentEntity NextItem = null; // entitys[k + 1];
//当前记录
CurrentItem = entitys[k];
//看看记录是不是已经存在了 如果存在则不进行更新
if (!docs.Contains(CurrentItem.Title.Trim()))
{
#region 值
//上一条记录
if (k - 1 >= 0)
{
PreviousItem = entitys[k - 1];
//如果上一章节已经存在了则就用数据库中的记录
var Temprecords = docentitys.Where(p => p.章节名.Trim() == PreviousItem.Title.Trim());
if (Temprecords.Count() > 0)
{
PreviousItem.Token = Temprecords.ElementAt(0).本记录GUID.ToString();
//设置本记录ID
CurrentItem.Token = Temprecords.ElementAt(0).一章.Value.ToString();
}
}
//下一条记录
if (k + 1 < entitys.Count)
{
NextItem = entitys[k + 1];
}
#endregion
//当前章节的GUID
if (CurrentItem.Token.Length < 10)
{
CurrentItem.Token = Guid.NewGuid().ToString();
}
//产生下一章节的GUID
if (NextItem != null && NextItem.Token.Length < 10)
{
NextItem.Token = Guid.NewGuid().ToString();
//如果上一章节已经存在了则就用数据库中的记录
var Temprecords = docentitys.Where(p => p.章节名.Trim() == CurrentItem.Title.Trim());
if (Temprecords.Count() > 0)
{
NextItem.Token = Temprecords.ElementAt(0).一章.Value.ToString();
}
}
//得到内容
string content = "";
try
{
content = documentAnalyse.GetContent(CurrentItem.Url.GetWeb());
}
catch
{
}
//添加到数据库
entities.AddTo文章表(new TygModel.文章表()
{
GUID = o.GUID,
书名 = o.书名,
分类标识 = o.分类标识,
本记录GUID = Guid.Parse(CurrentItem.Token),
创建时间 = DateTime.Now,
分类名称 = o.分类表.分类名称,
章节名 = CurrentItem.Title,
一章 = PreviousItem == null ? Guid.Empty :Guid.Parse(PreviousItem.Token),
一章 = NextItem == null ? Guid.Empty : Guid.Parse(NextItem.Token),
最后访问时间 = DateTime.Now,
内容 = content,
采集用的URL1 = CurrentItem.Url.ToString()
});
//打印状态
System.Diagnostics.Debug.WriteLine("采集内容:" + CurrentItem.Title + "成功,正文长度:" + content.Length);
}
else
{
//如果记录的内容比较少则更新内容
var Temprecords = docentitys.Where(p => p.章节名.Trim() == CurrentItem.Title.Trim());
if (Temprecords.Count() > 0)
{
//如果内容无效
if ( //中文小于6
string.Join("", System.Text.RegularExpressions.Regex.Matches(Temprecords.ElementAt(0).内容, @"[\u4e00-\u9fa5\d\w１２３４５６７８９～！!·＃￥％……—＊（）——＋／”》“‘’，；。、？，：…《]+[\u4e00-\u9fa5１２３４５６７８９～！!·＃￥％……—＊（!）——＋／”》“‘，’\r\n；。、？，：…《]", System.Text.RegularExpressions.RegexOptions.Multiline)
.Cast <System.Text.RegularExpressions.Match>().Select(p => p.Value).ToArray()
).Length < 6
)
{
//更新记录
var reccords = o.文章表.Where(p => p.章节名.Trim() == CurrentItem.Title.Trim());
if (reccords.Count() > 0)
{
var record = reccords.ElementAt(0);
try
{
record.采集用的URL1 = CurrentItem.Url.ToString();
record.内容 = documentAnalyse.GetContent(CurrentItem.Url.GetWeb());
}
catch { }
//打印状态
System.Diagnostics.Debug.WriteLine("更新记录:" + CurrentItem.Title + "成功,正文长度:" + record.内容.Length);
}
}
}
}
}
}
}
/// <summary>
/// 创建一个新的书本采集实例
/// </summary>
/// <param name="book">书</param>
public CreateBookHTML(TygModel.书名表 book)
{
Book = book;
//初始化进度文件
ProgressFileName = ProgressFileName + book.GUID + ".txt";
}
/// <summary>
/// 转换数据
/// </summary>
/// <returns></returns>
public override TygModel.书名表 Convert()
{
DateTime updateTime;
DateTime.TryParse(更新, out updateTime);
TygModel.书名表 book = null;
//看看分类表里有没有这个分类如果没有则进行分类添加
using (TygModel.Entities tygdb = new TygModel.Entities())
{
//分类表
var classItems = tygdb.分类表.Where(p => p.分类名称.Trim() == 类别.Trim());
//当前分类
TygModel.分类表 classItem = null;
//如果分类不存在
if (classItems.Count() == 0)
{
classItem = new TygModel.分类表()
{
分类标识 = 类别.Trim(),
分类名称 = 类别.Trim(),
分类说明 = 类别.Trim(),
备注 = "来自 yankuaikan.com",
通用分类 = 类别.Trim()
};
tygdb.AddTo分类表(classItem);
//保存分类
tygdb.SaveChanges();
}
else
{
classItem = classItems.FirstOrDefault();
}
book = new TygModel.书名表()
{
分类表 = classItem,
分类标识 = classItem.分类标识,
分类表ID = classItem.ID,
GUID = Guid.NewGuid(),
采集用的URL1 = 小说目录URL,
采集用的URL2 = 小说简介URL,
创建时间 = DateTime.Now,
最新章节 = 最新章节,
作者名称 = 作者,
说明 = "",
书名 = 小说名称.Replace("》", "").Replace("《", ""),
最后更新时间 = updateTime,
完本 = 状态.Trim() == "完成" ? true : false,
配图 = new Func <string>(() =>
{
if (小说简介URL != null)
{
//初始化一个DOM
HtmlAgilityPack.HtmlDocument dom = new HtmlAgilityPack.HtmlDocument();
dom.LoadHtml(小说简介URL.GetWeb());
//内容
HtmlAgilityPack.HtmlNode listContent = dom.GetElementbyId("content");
//可能的原素
List <PossiblyResultElement> possiblyResultElements = new List <PossiblyResultElement>();
//开始循环子原素
SingleListPageAnalyse.AnalyseMaxATagNearest(listContent, possiblyResultElements, 0, new PossiblyResultElement()
{
ParentPossiblyResult = null,
CurrnetHtmlElement = listContent,
LayerIndex = -1,
ContainTagNum = 0
});
//计算当前所有HTML原素中的img原素
var PageimgElements = from img in possiblyResultElements
where img.CurrnetHtmlElement.Name == "a" && img.CurrnetHtmlElement.HasChildNodes && img.CurrnetHtmlElement.ChildNodes.Where(p => p.Name == "img").Count() > 0
select img;
try
{
if (PageimgElements.Count() > 0)
{
//img.CurrnetHtmlElement.Attributes["src"].Value.Contains("http://tu.yankuai.com") && img.CurrnetHtmlElement.Attributes["src"] != null
string imgurl = PageimgElements.First().CurrnetHtmlElement.ChildNodes[0].Attributes["src"].Value;
return(imgurl.Trim().Contains("http://tu.yankuai.com") ? imgurl : "/images/noimg.jpg");
}
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine(DateTime.Now + ex.Message + "|||||" + ex.StackTrace);
}
}
//找到文章目录
return("/images/noimg.jpg");
}).Invoke(),
};
}
return(book);
}
/// <summary>
/// 转换数据
/// </summary>
/// <returns></returns>
public override TygModel.书名表 Convert()
{
DateTime updateTime = DateTime.Parse(更新);
//书本
TygModel.书名表 book = null;
//看看分类表里有没有这个分类如果没有则进行分类添加
using (TygModel.Entities tygdb = new TygModel.Entities())
{
//分类表
var classItems = tygdb.分类表.Where(p => p.分类名称.Trim() == 类别.Trim());
//当前分类
TygModel.分类表 classItem = null;
//如果分类不存在
if (classItems.Count() == 0)
{
try
{
classItem = new TygModel.分类表()
{
分类标识 = 类别.Trim(),
分类名称 = 类别.Trim(),
分类说明 = 类别.Trim(),
备注 = "来自 www.xs52.com",
通用分类 = 类别.Trim()
};
tygdb.AddTo分类表(classItem);
//保存分类
tygdb.SaveChanges();
}
catch (Exception ex)
{
System.Diagnostics.UDPGroup.SendStrGB2312(ex.Message + (ex.StackTrace != null ? ex.StackTrace : ""));
}
}
else
{
classItem = classItems.FirstOrDefault();
}
book = new TygModel.书名表()
{
分类标识 = classItem.分类标识,
分类表ID = classItem.ID,
GUID = Guid.NewGuid(),
采集用的URL1 = 小说目录URL,
采集用的URL2 = 小说简介URL,
创建时间 = DateTime.Now,
最新章节 = 最新章节,
作者名称 = 作者,
说明 = "",
书名 = 小说名称.Replace("》", "").Replace("《", ""),
最后更新时间 = updateTime,
完本 = 状态.Trim() == "连载" ? false : true,
配图 = "/images/noimg.gif",
};
//修改配图或者说明
if (小说简介URL != null)
{
book.说明 = string.Format("小说《{0}》{1}/著", 小说名称, 作者);
}
}
return(book);
}
/// <summary>
/// 创建一个新的书本采集实例
/// </summary>
/// <param name="book">书</param>
public CreateBookHTMLFTP(TygModel.书名表 book)
: base(book)
{
}
protected void Page_Load(object sender, EventArgs e)
{
string url = string.Format(BaseUrl, "1");
//得到最大的页码
int MaxPageNum = 1;
//初始化一个DOM
HtmlAgilityPack.HtmlDocument dom = new HtmlAgilityPack.HtmlDocument();
dom.LoadHtml(url.GetWeb());
//内容
HtmlAgilityPack.HtmlNode listContent = dom.GetElementbyId("content");
//得到分页
HtmlAgilityPack.HtmlNode pagesNode = dom.GetElementbyId("pagelink");
#region 得到页面的所有tr原素
List <BookInfoYankuaikan_com> books = GetBookInfos(listContent, url);
#endregion
#region 提取 pages 中 所有的a原素
MaxPageNum = GetMaxPageNum(pagesNode);
#endregion
List <BookInfoYankuaikan_com> AllBooks = new List <BookInfoYankuaikan_com>();
//序列化保存为文件
System.Xml.Serialization.XmlSerializer ser = new System.Xml.Serialization.XmlSerializer(AllBooks.GetType());
//如果文件不存在则从缓存读取
if (!System.IO.File.Exists(yankuankanPath))
{
AllBooks = GetBooks(MaxPageNum);//MaxPageNum
System.IO.StringWriter sw = new System.IO.StringWriter();
ser.Serialize(sw, AllBooks);
System.IO.File.WriteAllText(yankuankanPath, sw.ToString());
}
else
{
AllBooks = (List <BookInfoYankuaikan_com>)ser.Deserialize(new System.IO.StringReader(System.IO.File.ReadAllText(yankuankanPath)));
}
//System.Threading.Tasks.Task.Factory.StartNew(new Action(delegate
//{
//添加到Book中
for (int k = 0; k < AllBooks.Count; k++)
{
try
{
TygModel.书名表 book = AllBooks[k].Convert();
//当前记录
var records = Skybot.Cache.RecordsCacheManager.Instance.Tygdb.书名表.Where(p => p.书名.Trim() == book.书名.Trim() && p.作者名称.Trim() == book.作者名称.Trim());
if (records.Count() > 0)
{
//得到当前书更新记录
TygModel.书名表 currentBook = records.FirstOrDefault();
// currentBook.分类表 = book.分类表;
//currentBook.分类表ID = book.分类表ID;
//currentBook.GUID = book.GUID;
currentBook.采集用的URL1 = book.采集用的URL1;
currentBook.采集用的URL2 = book.采集用的URL2;
// currentBook.创建时间 = book.创建时间;
currentBook.最新章节 = book.最新章节;
//currentBook.作者名称 = book.作者名称;
currentBook.说明 = book.说明;
// currentBook.书名 = book.书名;
currentBook.最后更新时间 = book.最后更新时间;
currentBook.完本 = book.完本;
currentBook.配图 = book.配图;
//保存更改
Skybot.Cache.RecordsCacheManager.Instance.Tygdb.SaveChanges();
}
else
{
Skybot.Cache.RecordsCacheManager.Instance.Tygdb.AddTo书名表(book);
//保存更改
Skybot.Cache.RecordsCacheManager.Instance.Tygdb.SaveChanges();
}
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine(ex.Message + (ex.StackTrace != null ? ex.StackTrace : ""));
}
System.Diagnostics.Debug.WriteLine("已经完成书" + k + "/" + AllBooks.Count);
}
//}));
Response.Write("输入完成");
}
protected void Page_Load(object sender, EventArgs e)
{
Response.Write(
Server.UrlDecode("http://www.trimedgas.com/error/error.aspx?Error='s'+%e9%99%84%e8%bf%91%e6%9c%89%e8%af%ad%e6%b3%95%e9%94%99%e8%af%af%e3%80%82%0d%0a%e5%85%b3%e9%94%ae%e5%ad%97+'with'+%e9%99%84%e8%bf%91%e6%9c%89%e8%af%ad%e6%b3%95%e9%94%99%e8%af%af%e3%80%82%e5%a6%82%e6%9e%9c%e6%ad%a4%e8%af%ad%e5%8f%a5%e6%98%af%e5%85%ac%e7%94%a8%e8%a1%a8%e8%a1%a8%e8%be%be%e5%bc%8f%e6%88%96+xmlnamespaces+%e5%ad%90%e5%8f%a5%ef%bc%8c%e9%82%a3%e4%b9%88%e5%89%8d%e4%b8%80%e4%b8%aa%e8%af%ad%e5%8f%a5%e5%bf%85%e9%a1%bb%e4%bb%a5%e5%88%86%e5%8f%b7%e7%bb%93%e5%b0%be%e3%80%82%0d%0a%e5%85%b3%e9%94%ae%e5%ad%97+'with'+%e9%99%84%e8%bf%91%e6%9c%89%e8%af%ad%e6%b3%95%e9%94%99%e8%af%af%e3%80%82%e5%a6%82%e6%9e%9c%e6%ad%a4%e8%af%ad%e5%8f%a5%e6%98%af%e5%85%ac%e7%94%a8%e8%a1%a8%e8%a1%a8%e8%be%be%e5%bc%8f%e6%88%96+xmlnamespaces+%e5%ad%90%e5%8f%a5%ef%bc%8c%e9%82%a3%e4%b9%88%e5%89%8d%e4%b8%80%e4%b8%aa%e8%af%ad%e5%8f%a5%e5%bf%85%e9%a1%bb%e4%bb%a5%e5%88%86%e5%8f%b7%e7%bb%93%e5%b0%be%e3%80%82%0d%0a%e5%85%b3%e9%94%ae%e5%ad%97+'with'+%e9%99%84%e8%bf%91%e6%9c%89%e8%af%ad%e6%b3%95%e9%94%99%e8%af%af%e3%80%82%e5%a6%82%e6%9e%9c%e6%ad%a4%e8%af%ad%e5%8f%a5%e6%98%af%e5%85%ac%e7%94%a8%e8%a1%a8%e8%a1%a8%e8%be%be%e5%bc%8f%e6%88%96+xmlnamespaces+%e5%ad%90%e5%8f%a5%ef%bc%8c%e9%82%a3%e4%b9%88%e5%89%8d%e4%b8%80%e4%b8%aa%e8%af%ad%e5%8f%a5%e5%bf%85%e9%a1%bb%e4%bb%a5%e5%88%86%e5%8f%b7%e7%bb%93%e5%b0%be%e3%80%82"
)
);
var books = Skybot.Cache.RecordsCacheManager.Instance.Tygdb.书名表.ToList();
for (int classid = 1; classid <= 8; classid++)
{
System.Diagnostics.Debug.WriteLine("开始采集图书列表" + "http://www.86zw.com/Book/ShowBookList.aspx?tclassid=" + classid + "&page={0}");
var al = new Skybot.Collections.Sites.BookList86zw_com()
{
BaseUrl = "http://www.86zw.com/Book/ShowBookList.aspx?tclassid=" + classid + "&page={0}"
}.DoWork();
//开始写入数据库
var AllBooks = from doc in al select doc;
//添加到Book中
for (int k = 0; k < AllBooks.Count(); k++)
{
try
{
TygModel.书名表 book = AllBooks.ElementAt(k).Convert();
//当前记录
var records = books.Where(p => p.书名.Replace("》", "").Replace("《", "").Trim() == book.书名.Replace("》", "").Replace("《", "").Trim() && p.作者名称.Trim() == book.作者名称.Trim());
if (records.Count() > 0)
{
//得到当前书更新记录
TygModel.书名表 currentBook = records.FirstOrDefault();
// currentBook.分类表 = book.分类表;
//currentBook.分类表ID = book.分类表ID;
//currentBook.GUID = book.GUID;
currentBook.采集用的URL1 = book.采集用的URL1;
currentBook.采集用的URL2 = book.采集用的URL2;
// currentBook.创建时间 = book.创建时间;
currentBook.最新章节 = book.最新章节;
//currentBook.作者名称 = book.作者名称;
currentBook.说明 = book.说明;
// currentBook.书名 = book.书名;
currentBook.最后更新时间 = book.最后更新时间;
currentBook.完本 = book.完本;
currentBook.配图 = book.配图;
//保存更改
Skybot.Cache.RecordsCacheManager.Instance.Tygdb.SaveChanges();
}
else
{
Skybot.Cache.RecordsCacheManager.Instance.Tygdb.AddTo书名表(book);
Skybot.Cache.RecordsCacheManager.Instance.Tygdb.SaveChanges();
}
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine(ex.Message + (ex.StackTrace != null ? ex.StackTrace : ""));
}
System.Diagnostics.Debug.WriteLine("已经完成书" + k + "/" + AllBooks.Count());
}
}
System.Diagnostics.Debug.WriteLine("86zw 所有图书状态更新完成");
//更新书有效章节数量
TygModel.Entities enti = new TygModel.Entities();
var dd = enti.书名表.ToList();
dd = dd.Where(p => p.包含有效章节 == null || p.包含有效章节 == 0).ToList();
for (int k = 0; k < dd.Count(); k++)
{
dd.ElementAt(k).包含有效章节 = dd.ElementAt(k).文章表.Count;
enti.SaveChanges();
}
}
/// <summary>
/// 转换数据
/// </summary>
/// <returns></returns>
public override TygModel.书名表 Convert()
{
DateTime updateTime = DateTime.Parse("2000-01-01");
//书本
TygModel.书名表 book = null;
//看看分类表里有没有这个分类如果没有则进行分类添加
using (TygModel.Entities tygdb = new TygModel.Entities())
{
//分类表
var classItems = tygdb.分类表.Where(p => p.分类名称.Trim() == 类别.Trim());
//当前分类
TygModel.分类表 classItem = null;
//如果分类不存在
if (classItems.Count() == 0)
{
try
{
classItem = new TygModel.分类表()
{
分类标识 = 类别.Trim(),
分类名称 = 类别.Trim(),
分类说明 = 类别.Trim(),
备注 = "来自 86zw.com",
通用分类 = 类别.Trim()
};
tygdb.AddTo分类表(classItem);
//保存分类
tygdb.SaveChanges();
}
catch (Exception ex)
{
System.Diagnostics.UDPGroup.SendStrGB2312(ex.Message + (ex.StackTrace != null ? ex.StackTrace : ""));
}
}
else
{
classItem = classItems.FirstOrDefault();
}
book = new TygModel.书名表()
{
分类表 = classItem,
分类标识 = classItem.分类标识,
分类表ID = classItem.ID,
GUID = Guid.NewGuid(),
采集用的URL1 = 小说目录URL,
采集用的URL2 = 小说简介URL,
创建时间 = DateTime.Now,
最新章节 = 最新章节,
作者名称 = 作者,
说明 = "",
书名 = 小说名称.Replace("》", "").Replace("《", ""),
最后更新时间 = updateTime,
完本 = 状态.Trim() == "完结" ? true : false,
配图 = "/images/noimg.gif",
};
//修改配图或者说明
if (小说简介URL != null)
{
System.Diagnostics.UDPGroup.SendStrGB2312("获取配图：" + 小说简介URL);
//初始化一个DOM
HtmlAgilityPack.HtmlDocument dom = new HtmlAgilityPack.HtmlDocument();
dom.LoadHtml(小说简介URL.GetWeb());
//信息说明字段
HtmlAgilityPack.HtmlNode DesriptionContent = dom.GetElementbyId("CrbtrTop");
//采集时间的xpath表达式 2011-11-18
HtmlAgilityPack.HtmlNode node = dom.DocumentNode.SelectSingleNode("/html[1]/body[1]/div[2]/div[2]/div[3]/div[2]/div[1]/ul[1]/li[6]");
if (node != null)
{
if (DateTime.TryParse(node.InnerText, out updateTime))
{
book.最后更新时间 = updateTime;
}
}
//说明
HtmlAgilityPack.HtmlNode summary = dom.GetElementbyId("CrbsSum");
if (summary != null)
{
book.说明 = summary.InnerHtml.Length > 4000 ? new Tong.TongUse().ForMatText(summary.InnerText, 0, 3800) : summary.InnerHtml;
}
//图片
HtmlAgilityPack.HtmlNode listContent = dom.GetElementbyId("CrbtlBookImg");
if (listContent != null)
{
//可能的原素
List <PossiblyResultElement> possiblyResultElements = new List <PossiblyResultElement>();
//开始循环子原素
SingleListPageAnalyse.AnalyseMaxATagNearest(listContent, possiblyResultElements, 0, new PossiblyResultElement()
{
ParentPossiblyResult = null,
CurrnetHtmlElement = listContent,
LayerIndex = -1,
ContainTagNum = 0
});
//计算当前所有HTML原素中的img原素
var PageimgElements = from img in possiblyResultElements
where img.CurrnetHtmlElement.Name == "img"
select img;
try
{
if (PageimgElements.Count() > 0)
{
string imgurl = PageimgElements.First().CurrnetHtmlElement.Attributes["src"].Value;
if (!imgurl.ToLower().Contains("images/noimg.gif"))
{
try
{
imgurl = new Uri(new Uri(小说简介URL), imgurl).ToString();
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine(DateTime.Now + ex.Message + "|||||" + ex.StackTrace);
imgurl = "/images/noimg.gif";
}
}
System.Diagnostics.UDPGroup.SendStrGB2312("获取配图：" + 小说简介URL + " 完成" + imgurl);
book.配图 = imgurl.Trim().Contains("/images/noimg.gif") ? "/images/noimg.gif" : imgurl;
}
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine(DateTime.Now + ex.Message + "|||||" + ex.StackTrace);
}
}
}
}
return(book);
}
