public override bool Download()
{
CurrentParameter.IsStop = false;
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
new BrRegex(),
new HtmlDecode(),
new UniformFormat(),
new Traditional()
};
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
string url = string.Format("http://www.ranwen.net/files/article/{0}/{1}/{2}.html",
(CommonTools.TryParse(TaskInfo.Tid, 0) / 1000).ToString(CultureInfo.InvariantCulture),
TaskInfo.Tid,
SectionNames[TaskInfo.CurrentSection].ToString(CultureInfo.InvariantCulture));//組合網址
HtmlDocument htmlRoot = GetHtmlDocument(url);
try
{
string tempTextFile = htmlRoot.DocumentNode.SelectSingleNode("//*[@id=\"bgdiv\"]/table[2]/tbody/tr[1]/td/div[1]/h1").InnerText
+ "\r\n";
var node = htmlRoot.DocumentNode.SelectSingleNode("//*[@id=\"content\"]");
Network.RemoveSubHtmlNode(node, "div");
tempTextFile += node.InnerHtml + "\r\n";
foreach (var item in typeSetting)
{
item.Set(ref tempTextFile);
}
FileWrite.TxtWrire(tempTextFile, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}
public override bool Download()
{
CurrentParameter.IsStop = false;
string urlHead = string.Format("http://book.sfacg.com/Novel/{0}", TaskInfo.Tid);
//string urlTail = ".html?charset=big5";
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
new SfacgToIndent(),
new HtmlDecode(),
new UniformFormat(),
new Traditional()
};
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
string url = urlHead + SectionNames[TaskInfo.CurrentSection].ToString(CultureInfo.InvariantCulture);//組合網址
HtmlDocument htmlRoot = GetHtmlDocument(url);
try
{
var nodeHeaders =
htmlRoot.DocumentNode.SelectSingleNode(@"//*[@id=""ChapterBody""]");
Network.RemoveSubHtmlNode(nodeHeaders, "img");
string tempTextFile = nodeHeaders.InnerHtml;
foreach (var item in typeSetting)
{
item.Set(ref tempTextFile);
}
FileWrite.TxtWrire(tempTextFile, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}
public override bool Download()
{
CurrentParameter.IsStop = false;
string urlHead = "http://lknovel.lightnovel.cn/main/view/";
string urlTail = ".html?charset=big5";
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
new HtmlDecode(),
new UniformFormat()
};
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
string url = urlHead + SectionNames[TaskInfo.CurrentSection].ToString(CultureInfo.InvariantCulture) + urlTail;//組合網址
HtmlDocument htmlRoot = GetHtmlDocument(url);
try
{
var nodeHeaders =
htmlRoot.DocumentNode.SelectSingleNode(@"//*[@id=""J_view""]");
string tempTextFile = nodeHeaders.InnerText;
foreach (var item in typeSetting)
{
item.Set(ref tempTextFile);
}
FileWrite.TxtWrire(tempTextFile, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}
public override bool Download()
{
CurrentParameter.IsStop = false;
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
new BrRegex(),
new HtmlDecode(),
new UniformFormat(),
new Traditional()
};
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
try
{
string url = string.Format("http://www.biquge.com/{0}/{1}.html", TaskInfo.Tid, SectionNames[TaskInfo.CurrentSection].ToString(CultureInfo.InvariantCulture));//組合網址
HtmlDocument htmlRoot = GetHtmlDocument(url);
string tempTextFile = htmlRoot.DocumentNode.SelectSingleNode("//*[@id=\"wrapper\"]/div[3]/div[1]/div[2]/h1").InnerText
+ "\r\n" + htmlRoot.DocumentNode.SelectSingleNode("//*[@id=\"content\"]").InnerHtml + "\r\n";
foreach (var item in typeSetting)
{
item.Set(ref tempTextFile);
}
FileWrite.TxtWrire(tempTextFile, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}
public override bool Download()
{
CurrentParameter.IsStop = false;
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
new HtmlDecode(),
new UniformFormat(),
new HjwzwRegex()
};
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
try
{
string url = string.Format("http://tw.hjwzw.com/Book/Read/{0},{1}",
TaskInfo.Tid,
SectionNames[TaskInfo.CurrentSection].ToString(CultureInfo.InvariantCulture));//組合網址
HtmlDocument htmlRoot = GetHtmlDocument(url);
string tempTextFile = htmlRoot.DocumentNode.SelectSingleNode("/html/body/table[7]/tr/td/div[5]").InnerText;
foreach (var item in typeSetting)
{
item.Set(ref tempTextFile);
}
FileWrite.TxtWrire(tempTextFile, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}
public override bool Download()
{
CurrentParameter.IsStop = false;
Regex r = new Regex(@"(?<Head>^https?:\/\/\w*\.*ck101.com\/thread-\d+-)(?<CurrentPage>\d+)(?<Tail>-\w+\.html)");
Match m = r.Match(TaskInfo.Url);
string urlHead = string.Empty, urlTail = string.Empty;
if (m.Success)
{
urlHead = m.Groups["Head"].Value;
urlTail = m.Groups["Tail"].Value;
}
HtmlNodeCollection nodeHeaders = null;
int lastPage = 0;
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
new HtmlDecode(),
new UniformFormat()
};
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
//要下載的頁數
try
{
int newCurrentPage = (TaskInfo.BeginSection + TaskInfo.PageSection - 1) / TaskInfo.PageSection;
if (lastPage != newCurrentPage) //之前下載的頁數跟當前要下載的頁數
{
lastPage = newCurrentPage; //記錄下載頁數，下次如果一樣就不用重抓
string url = urlHead + lastPage.ToString(CultureInfo.InvariantCulture) + urlTail; //組合網址
if (lastPage == 1) //卡提諾第一頁的特別處理
{
switch (TaskInfo.FailTimes % 2) //常常取不到完整資料，用多個網址取
{
case 0:
url = string.Format("https://ck101.com/thread-{0}-1-1.html", TaskInfo.Tid);
break;
case 1:
url = string.Format("https://m.ck101.com/forum.php?mod=redirect&ptid={0}&authorid=0&postno=1", TaskInfo.Tid);
break;
case 2:
url = string.Format("https://m.ck101.com/forum.php?mod=redirect&ptid={0}&authorid=0&postno=1", TaskInfo.Tid);
break;
}
}
HtmlDocument htmlRoot = GetHtmlDocumentReplaceDivToEmpty(url);
if (htmlRoot != null)
{
nodeHeaders = htmlRoot.DocumentNode.SelectNodes("//*[@class=\"t_f\"]");
}
}
//計算要取的區塊在第幾個
int partSection = TaskInfo.BeginSection - ((lastPage - 1) * TaskInfo.PageSection) - 1;
if (nodeHeaders == null)
{
throw new Exception("下載資料為空的");
}
Network.RemoveSubHtmlNode(nodeHeaders[partSection], "div");
Network.RemoveSubHtmlNode(nodeHeaders[partSection], "ignore_js_op");
Network.RemoveSubHtmlNode(nodeHeaders[partSection], "i");
Network.RemoveSubHtmlNode(nodeHeaders[partSection], "script");
string tempTxt = nodeHeaders[partSection].InnerText;
foreach (var item in typeSetting)
{
item.Set(ref tempTxt);
}
FileWrite.TxtWrire(tempTxt, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
lastPage = 0;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}
public override bool Download()
{
CurrentParameter.IsStop = false;
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
new AnnotationRegex(),
new BrRegex(),
new PRegex(),
new HtmlDecode(),
new UniformFormat(),
};
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
string url = string.Format("http://big5.quanben5.com/n/{0}/{1}.html",
TaskInfo.Tid,
SectionNames[TaskInfo.CurrentSection].ToString(CultureInfo.InvariantCulture));//組合網址
HtmlDocument htmlRoot = GetHtmlDocument(url);
try
{
Regex r = new Regex(@"ajax_post\('book','ajax_content','pinyin','(?<pinyin>\S+)','content_id','(?<content_id>\d+)','sky','(?<sky>\S+)','t','(?<t>\d+)'\)");
Match m = r.Match(htmlRoot.DocumentNode.InnerHtml);
string formData = "";
if (m.Success)
{
string timestamp = Convert.ToInt32(DateTime.UtcNow.AddHours(8).Subtract(new DateTime(1970, 1, 1)).TotalSeconds).ToString() + "000";
formData = string.Format("pinyin={0}&content_id={1}&sky={2}&t={3}&_type=ajax&rndval={4}",
m.Groups["pinyin"].Value,
m.Groups["content_id"].Value,
m.Groups["sky"].Value,
m.Groups["t"].Value,
timestamp
);
}
HtmlDocument htmlPostResponse = PostHtmlDocument(formData);
var titleNode = htmlRoot.DocumentNode.SelectSingleNode("/html/body/div[3]/div/div[2]/h1");
//Network.RemoveSubHtmlNode(node, "div");
string tempTextFile = titleNode.InnerText + "\r\n" + htmlPostResponse.DocumentNode.InnerHtml + "\r\n";
foreach (var item in typeSetting)
{
item.Set(ref tempTextFile);
}
FileWrite.TxtWrire(tempTextFile, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}
public override bool Download()
{
CurrentParameter.IsStop = false;
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
//new AnnotationRegex(),
new Remove0007(),
new BrRegex(),
//new PRegex(),
new HtmlDecode(),
new UniformFormat(),
};
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
string url = string.Format("https://www.wfxs.org/html/{0}/{1}.html",
TaskInfo.Tid,
SectionNames[TaskInfo.CurrentSection].ToString(CultureInfo.InvariantCulture));//組合網址
try
{
string htmlstring = GetHtmlString(url);
string content = "";
Regex r = new Regex(@"<a href=""\/html\/\d+\/"">.+?<\/a>(?<content>.+?)<script>style_4\(\);<\/script>", RegexOptions.Singleline);
Match m = r.Match(htmlstring);
if (m.Success)
{
content = m.Groups["content"].Value;
}
HtmlDocument htmlRoot = Network.GetHtmlDocument(content);
var node = htmlRoot.DocumentNode;
Network.RemoveSubHtmlNode(node, "div");
string tempTextFile = node.InnerHtml + "\r\n";
foreach (var item in typeSetting)
{
item.Set(ref tempTextFile);
}
FileWrite.TxtWrire(tempTextFile, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}
public override bool Download()
{
//Regex r = new Regex(@"(?<Head>^http:\/\/\w*\.eyny.com\/thread-\d+-)(?<CurrentPage>\d+)(?<Tail>-\w+\.html)");
//Match m = r.Match(TaskInfo.Url);
string urlHead = string.Empty, urlTail = string.Empty;
//if (m.Success)
//{
// urlHead = m.Groups["Head"].Value;
// urlTail = m.Groups["Tail"].Value;
//}
//http://archiver.eyny.com/archiver/tid-9169460-1.html
urlHead = string.Format(@"http://archiver.eyny.com/archiver/tid-{0}-", TaskInfo.Tid);
urlTail = @".html";
HtmlNodeCollection nodeHeaders = null;
int lastPage = 0;
//排版插件
var typeSetting = new Collection <ITypeSetting>
{
new HtmlDecode(),
new EynyTag(),
new UniformFormat()
};
string RawData = "";
for (; TaskInfo.BeginSection <= TaskInfo.EndSection && !CurrentParameter.IsStop; TaskInfo.BeginSection++)
{
try
{
//要下載的頁數
int newCurrentPage = (TaskInfo.BeginSection + TaskInfo.PageSection - 1) / TaskInfo.PageSection;
if (lastPage != newCurrentPage) //之前下載的頁數跟當前要下載的頁數
{
lastPage = newCurrentPage; //記錄下載頁數，下次如果一樣就不用重抓
string url = urlHead + lastPage + urlTail; //組合網址
HtmlDocument htmlRoot = GetHtmlDocument(url);
if (htmlRoot != null)
{
nodeHeaders = htmlRoot.DocumentNode.SelectNodes("//*[@id=\"content\"]");
}
Network.RemoveSubHtmlNode(nodeHeaders[0], "div");
Network.RemoveSubHtmlNode(nodeHeaders[0], "ignore_js_op");
Network.RemoveSubHtmlNode(nodeHeaders[0], "i");
Network.RemoveSubHtmlNode(nodeHeaders[0], "p", "strong");
RawData = nodeHeaders[0].InnerText;
RawData += "\r\n發表於 2001-1-1 1:1 PM";
foreach (var item in typeSetting)
{
item.Set(ref RawData);
}
if (nodeHeaders == null)
{
throw new Exception("下載資料為空的");
}
}
//Network.RemoveSubHtmlNode(nodeHeaders[0], "p");
//計算要取的區塊在第幾個
int partSection = TaskInfo.BeginSection - ((lastPage - 1) * TaskInfo.PageSection) - 1;
Regex r = new Regex(@"((發表於(( [昨前]天 \d+:\d+ [PA]M)|( \d+-\d+-\d+ \d+:\d+ [PA]M)|( .+?前))))(?<Main>.+?)(?=(發表於(( [昨前]天 \d+:\d+ [PA]M)|( \d+-\d+-\d+ \d+:\d+ [PA]M)|( .+?前))))", RegexOptions.Singleline);
var m = r.Matches(RawData);
string tempTxt = m[partSection].Groups["Main"].Value;
FileWrite.TxtWrire(tempTxt, TaskInfo.SaveFullPath, TaskInfo.TextEncoding);
}
catch (Exception)
{
//CoreManager.LoggingManager.Debug(ex.ToString());
//發生錯誤，當前區塊重取
TaskInfo.BeginSection--;
TaskInfo.FailTimes++;
lastPage = 0;
continue;
}
TaskInfo.HasStopped = CurrentParameter.IsStop;
}
bool finish = TaskInfo.CurrentSection == TaskInfo.EndSection;
return(finish);
}