private IEnumerator <RapideDdl_PostHeader> _Load(string url, bool reload, bool loadImage)
{
if (url != null)
{
// dont use mongo to store header page so key is null and refreshDocumentStore is false
RequestWebData_v3 request = new RequestWebData_v3(new RequestFromWeb_v3(url, _requestParameters, reload, loadImage));
_headerPage = RapideDdl_LoadHeaderPage.Load.Load(request).Document;
if (loadImage)
{
foreach (var postHeader in _headerPage.postHeaders)
{
postHeader.images = DownloadPrint.LoadImages(postHeader.images).ToArray();
}
}
if (_headerPage != null)
{
return(_headerPage.postHeaders.AsEnumerable <RapideDdl_PostHeader>().GetEnumerator());
}
}
return(null);
}
// detail get data
protected override MagazinesGratuits_PostDetail GetDetailData(WebResult webResult)
{
XXElement xeSource = webResult.Http.zGetXDocument().zXXElement();
MagazinesGratuits_PostDetail data = new MagazinesGratuits_PostDetail();
data.SourceUrl = webResult.WebRequest.HttpRequest.Url;
data.LoadFromWebDate = webResult.LoadFromWebDate;
data.Key = _GetDetailKey(webResult.WebRequest.HttpRequest);
XXElement xePost = xeSource.XPathElement("//table[@id='layout']//div[@id='content']//div[@class='post']");
XXElement xe = xePost.XPathElement(".//table[@id='post-head']");
string[] dates = xe.XPathElement(".//td[@id='head-date']").DescendantTexts().Select(DownloadPrint.Trim).ToArray();
data.PostCreationDate = GetDate(dates, _lastPostDate);
if (data.PostCreationDate != null)
{
_lastPostDate = new Date(data.PostCreationDate.Value);
}
if (__trace)
{
pb.Trace.WriteLine("post creation date {0} - {1}", data.PostCreationDate, dates.zToStringValues());
}
data.Title = xePost.XPathValue(".//div[@class='title']//a//text()").zFunc(DownloadPrint.ReplaceChars).zFunc(DownloadPrint.Trim);
PrintTitleInfos titleInfos = DownloadPrint.PrintTextValuesManager.ExtractTitleInfos(data.Title);
if (titleInfos.foundInfo)
{
data.OriginalTitle = data.Title;
data.Title = titleInfos.title;
data.Infos.SetValues(titleInfos.infos);
}
// Ebooks en Epub / Livre
data.Category = xePost.XPathElements(".//div[@class='postdata']//span[@class='category']//a").DescendantTexts().Select(DownloadPrint.Trim).zToStringValues("/");
data.PrintType = GetPrintType(data.Category);
//pb.Trace.WriteLine("category \"{0}\" printType {1}", category, data.printType);
xe = xePost.XPathElement(".//div[@class='entry']");
data.Images = new WebImage[] { new WebImage(zurl.GetUrl(data.SourceUrl, xe.XPathValue("div[starts-with(@class, 'post-views')]/following-sibling::h3/following-sibling::p/img/@src"))) };
// force load image to get image width and height
if (webResult.WebRequest.LoadImage)
{
data.Images = DownloadPrint.LoadImages(data.Images).ToArray();
}
// get infos, description, language, size, nbPages
// xe.DescendantTextList(".//p")
PrintTextValues textValues = DownloadPrint.PrintTextValuesManager.GetTextValues(
xe.XPathElements(".//p").DescendantTexts(
node =>
{
if (node is XText)
{
string text = ((XText)node).Value.Trim();
//if (text.StartsWith("Lien Direct", StringComparison.InvariantCultureIgnoreCase))
if (text.StartsWith("lien ", StringComparison.InvariantCultureIgnoreCase))
{
return(XNodeFilter.Stop);
}
}
if (node is XElement)
{
XElement xe2 = (XElement)node;
if (xe2.Name == "p" && xe2.zAttribValue("class") == "submeta")
{
return(XNodeFilter.Stop);
}
}
return(XNodeFilter.SelectNode);
}
).Select(DownloadPrint.ReplaceChars).Select(DownloadPrint.TrimWithoutColon), data.Title);
data.Description = textValues.description;
//data.Language = textValues.language;
//data.Size = textValues.size;
//data.NbPages = textValues.nbPages;
data.Infos.SetValues(textValues.infos);
//data.DownloadLinks = xe.DescendantNodes(
// node =>
// {
// if (!(node is XElement))
// return true;
// XElement xe2 = (XElement)node;
// if (xe2.Name != "p")
// return true;
// XAttribute xa = xe2.Attribute("class");
// if (xa == null)
// return true;
// if (xa.Value != "submeta")
// return true;
// return false;
// },
// node => node is XElement && ((XElement)node).Name == "a")
// .Select(node => ((XElement)node).Attribute("href").Value).ToArray();
data.DownloadLinks = xe.DescendantNodes(
node =>
{
if (!(node is XElement))
{
return(XNodeFilter.DontSelectNode);
}
XElement xe2 = (XElement)node;
if (xe2.Name == "a")
{
return(XNodeFilter.SelectNode);
}
if (xe2.Name != "p")
{
return(XNodeFilter.DontSelectNode);
}
XAttribute xa = xe2.Attribute("class");
if (xa == null)
{
return(XNodeFilter.DontSelectNode);
}
if (xa.Value != "submeta")
{
return(XNodeFilter.DontSelectNode);
}
//return XNodeFilter.SkipNode;
return(XNodeFilter.Stop);
})
.Select(node => ((XElement)node).Attribute("href").Value).ToArray();
//// <div id="postlist" class="postlist restrain">
//xe = xePost.XPathElement(".//div[@id='postlist']");
//// Aujourd'hui, 07h32 - Aujourd'hui, 10h51 - Hier, 12h55 - 22/02/2014, 21h09
////string date = xe.DescendantTextList(".//div[@class='posthead']//text()", nodeFilter: node => node.zGetName() != "a").zToStringValues("");
//XXElement xe2 = xe.XPathElement(".//div[@class='posthead']");
//string date = xe2.DescendantTextList(nodeFilter: node => node.zGetName() != "a").zToStringValues("");
//date = date.Replace('\xA0', ' ');
//data.PostCreationDate = zdate.ParseDateTimeLikeToday(date, webResult.LoadFromWebDate, @"d/M/yyyy, HH\hmm", @"d-M-yyyy, HH\hmm");
//if (data.PostCreationDate == null)
// pb.Trace.WriteLine("unknow post creation date \"{0}\"", date);
//data.PostAuthor = xe.XPathValue(".//div[@class='userinfo']//a//text()", DownloadPrint.TrimFunc1);
//// <div class="postbody">
//xe = xePost.XPathElement(".//div[@class='postbody']//div[@class='content']//blockquote/div");
//data.Images = xe.XPathImages(xeImg => new UrlImage(zurl.GetUrl(data.SourceUrl, xeImg.zAttribValue("src")))).ToArray();
//// get infos, description, language, size, nbPages
//PrintTextValues textValues = DownloadPrint.PrintTextValuesManager.GetTextValues(xe.DescendantTextList(nodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a"), data.Title);
//data.Description = textValues.description;
//data.Language = textValues.language;
//data.Size = textValues.size;
//data.NbPages = textValues.nbPages;
//data.Infos.SetValues(textValues.infos);
//data.DownloadLinks = xe.XPathValues(".//a/@href");
if (__trace)
{
pb.Trace.WriteLine(data.zToJson());
}
return(data);
}
private static ExtremeDown_PostDetail_v2 GetData(WebResult webResult)
{
XXElement xeSource = webResult.Http.zGetXDocument().zXXElement();
ExtremeDown_PostDetail_v2 data = new ExtremeDown_PostDetail_v2();
data.SourceUrl = webResult.WebRequest.HttpRequest.Url;
data.LoadFromWebDate = webResult.LoadFromWebDate;
data.Id = GetPostDetailKey(webResult.WebRequest.HttpRequest);
XXElement xePost = xeSource.XPathElement("//div[@id='dle-content']");
//data.Title = xePost.XPathValue(".//h2[@class='blocktitle']//text()", DownloadPrint.Trim);
data.Title = xePost.XPathValue(".//h2[@class='blocktitle']//text()").Trim(DownloadPrint.TrimChars);
PrintTitleInfos titleInfos = DownloadPrint.PrintTextValuesManager.ExtractTitleInfos(data.Title);
if (titleInfos.foundInfo)
{
data.OriginalTitle = data.Title;
data.Title = titleInfos.title;
data.Infos.SetValues(titleInfos.infos);
}
XXElement xeDiv = xePost.XPathElement(".//div[@class='blockheader']");
data.Category = xeDiv.XPathValues(".//i[@class='icon-cats']/ancestor::span//a//text()").Select(DownloadPrint.Trim).zToStringValues("/");
//string category = data.Category.ToLowerInvariant();
data.PrintType = GetPrintType(data.Category);
data.PostAuthor = xeDiv.XPathValue(".//span/i[@class='icon-user']/ancestor::span//a//text()");
string date = xeDiv.XPathValue(".//span/i[@class='icon-date']/ancestor::span//a//text()");
data.PostCreationDate = zdate.ParseDateTimeLikeToday(date, webResult.LoadFromWebDate, "d-M-yyyy, HH:mm", "d M yyyy", "d MMMM yyyy");
if (data.PostCreationDate == null)
{
pb.Trace.WriteLine("unknow date time \"{0}\"", date);
}
if (__trace)
{
pb.Trace.WriteLine("creationDate {0} - \"{1}\"", data.PostCreationDate, date);
}
xeDiv = xePost.XPathElement(".//div[@class='blockcontent']");
List <string> description = new List <string>();
description.AddRange(xeDiv.XPathValues(".//p[@class='release-name']//text()"));
//data.Images = xeDiv.XPathElement(".//table//td[@class='image-block']").XPathImages(xeImg => new UrlImage(zurl.GetUrl(loadDataFromWeb.request.Url, xeImg.zAttribValue("src")))).ToArray();
data.Images = xeDiv.XPathElement(".//table//td[@class='image-block']").DescendantNodes(node => XmlDescendant.ImageFilter(node)).Select(xeImg => new WebImage(zurl.GetUrl(data.SourceUrl, xeImg.zAttribValue("src")))).ToArray();
// force load image to get image width and height
if (webResult.WebRequest.LoadImage)
{
data.Images = DownloadPrint.LoadImages(data.Images).ToArray();
}
description.AddRange(xeDiv.XPathValues(".//table//td//blockquote//text()"));
//xeDiv = xePost.XPathElement(".//div[@class='clearfix']");
xeDiv = xePost.XPathElement(".//div[@class='upload-infos clearfix']");
description.AddRange(xeDiv.XPathValues(".//table//text()"));
data.Description = description.ToArray();
string title = null;
// xePost.XPathElements(".//script/parent::div//following-sibling::h2")
Func <XXElement, XNodeFilter> filter =
xe =>
{
if (xe.XElement.Name == "h2")
{
title = xe.XPathValue(".//text()");
}
else if (xe.XElement.Name == "script")
{
return(XNodeFilter.Stop);
}
else if (xe.XElement.Name == "div")
{
return(XNodeFilter.SelectNode);
}
return(XNodeFilter.DontSelectNode);
};
foreach (XXElement xe in xePost.XPathElements(".//div[@class='prez_2']//following-sibling::*").zFilterElements(filter))
{
//string s = xe.XPathValue(".//text()");
//// Liens de téléchargement - Pack 1
//if (s.StartsWith("Liens de téléchargement"))
//{
// s = s.Substring(23).Trim(' ', '-');
// if (s == "")
// s = title;
// else if (title != null)
// s = title + " - " + s;
// title = null;
data.DownloadLinks_new.AddItem(title);
title = null;
//foreach (XXElement xe2 in xe.XPathElements("following-sibling::div[1]//a"))
foreach (XXElement xe2 in xe.XPathElements(".//a"))
{
//s = xe2.DescendantTextList().FirstOrDefault();
// <strong class="hebergeur">
string server = xe2.XPathValue(".//strong[@class='hebergeur']//text()");
string link = xe2.XPathValue("@href");
if (__getLinksExtremeProtect && __extremeProtect.IsLinkProtected(link))
{
data.DownloadLinks_new.AddServer(server, link);
data.DownloadLinks_new.AddLinks(__extremeProtect.UnprotectLink(link));
}
else
{
data.DownloadLinks_new.AddServer(server);
data.DownloadLinks_new.AddLink(link);
}
}
//}
//else if (s != null)
// title = s;
}
//xeDiv = xePost.XPathElement(".//div[@class='blockfooter links']");
////data.category = xeDiv.DescendantTextList(".//i[@class='icon-cats']/parent::span//a").Select(DownloadPrint.TrimFunc1).zToStringValues("/");
//data.Category = xeDiv.XPathElements(".//i[@class='icon-cats']/parent::span//a").DescendantTexts().Select(DownloadPrint.Trim).zToStringValues("/");
//string category = data.Category.ToLowerInvariant();
//data.PrintType = GetPrintType(category);
////pb.Trace.WriteLine("category \"{0}\" printType {1}", category, data.printType);
if (__trace)
{
pb.Trace.WriteLine(data.zToJson());
}
return(data);
}
protected override IPost GetData(LoadDataFromWeb_v4 loadDataFromWeb)
{
XXElement xeSource = new XXElement(loadDataFromWeb.Http.zGetXDocument().Root);
Ebookdz_PostDetail data = new Ebookdz_PostDetail();
data.SourceUrl = loadDataFromWeb.WebRequest.HttpRequest.Url;
data.LoadFromWebDate = loadDataFromWeb.LoadFromWebDate;
data.Id = GetPostDetailKey(loadDataFromWeb.WebRequest.HttpRequest);
// <div class="body_bd">
XXElement xePost = xeSource.XPathElement("//div[@class='body_bd']");
// Le Monde + Magazine + 2 suppléments du samedi 03 janvier 2015
//data.Title = xePost.XPathValue(".//div[@id='pagetitle']//a//text()", DownloadPrint.Trim);
data.Title = xePost.XPathValue(".//div[@id='pagetitle']//a//text()").Trim(DownloadPrint.TrimChars);
PrintTitleInfos titleInfos = DownloadPrint.PrintTextValuesManager.ExtractTitleInfos(data.Title);
if (titleInfos.foundInfo)
{
data.OriginalTitle = data.Title;
data.Title = titleInfos.title;
data.Infos.SetValues(titleInfos.infos);
}
// Forum / Journaux / Presse quotidienne / Le Monde / Journal Le Monde + Magazine + 2 suppléments du samedi 03 janvier 2015
string lowerTitle = null;
if (data.Title != null)
{
lowerTitle = data.Title.ToLowerInvariant();
}
//data.Category = xePost.DescendantTextList(".//div[@id='breadcrumb']//a").Where(text => { text = text.ToLowerInvariant(); return text != "forum" && !text.EndsWith(lowerTitle); }).Select(DownloadPrint.TrimFunc1).zToStringValues("/");
data.Category = xePost.XPathElements(".//div[@id='breadcrumb']//a").DescendantTexts().Where(text => { text = text.ToLowerInvariant(); return(text != "forum" && !text.EndsWith(lowerTitle)); }).Select(DownloadPrint.Trim).zToStringValues("/");
string category = data.Category.ToLowerInvariant();
data.PrintType = GetPrintType(category);
//Trace.WriteLine("category \"{0}\" printType {1}", category, data.printType);
// <div id="postlist" class="postlist restrain">
XXElement xe = xePost.XPathElement(".//div[@id='postlist']");
// Aujourd'hui, 07h32 - Aujourd'hui, 10h51 - Hier, 12h55 - 22/02/2014, 21h09
//string date = xe.DescendantTextList(".//div[@class='posthead']//text()", nodeFilter: node => node.zGetName() != "a").zToStringValues("");
XXElement xe2 = xe.XPathElement(".//div[@class='posthead']");
//string date = xe2.DescendantTextList(nodeFilter: node => node.zGetName() != "a").zToStringValues("");
string date = xe2.DescendantTexts(node => node.zGetName() != "a" ? XNodeFilter.SelectNode : XNodeFilter.SkipNode).zToStringValues("");
date = date.Replace('\xA0', ' ');
data.PostCreationDate = zdate.ParseDateTimeLikeToday(date, loadDataFromWeb.LoadFromWebDate, @"d/M/yyyy, HH\hmm", @"d-M-yyyy, HH\hmm");
if (data.PostCreationDate == null)
{
pb.Trace.WriteLine("unknow post creation date \"{0}\"", date);
}
if (__trace)
{
pb.Trace.WriteLine("post creation date {0} - \"{1}\"", data.PostCreationDate, date);
}
//data.PostAuthor = xe.XPathValue(".//div[@class='userinfo']//a//text()", DownloadPrint.Trim);
data.PostAuthor = xe.XPathValue(".//div[@class='userinfo']//a//text()").Trim(DownloadPrint.TrimChars);
// <div class="postbody">
xe = xePost.XPathElement(".//div[@class='postbody']//div[@class='content']//blockquote/div");
//data.Images = xe.XPathImages(xeImg => new UrlImage(zurl.GetUrl(data.SourceUrl, xeImg.zAttribValue("src")))).ToArray();
data.Images = xe.DescendantNodes(node => XmlDescendant.ImageFilter(node)).Select(xeImg => new WebImage(zurl.GetUrl(data.SourceUrl, xeImg.zAttribValue("src")))).ToArray();
// force load image to get image width and height
if (loadDataFromWeb.WebRequest.LoadImage)
{
data.Images = DownloadPrint.LoadImages(data.Images).ToArray();
}
// get infos, description, language, size, nbPages
// xe.DescendantTextList(nodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a")
PrintTextValues_v1 textValues = DownloadPrint.PrintTextValuesManager.GetTextValues_v1(xe.DescendantTexts(node => !(node is XElement) || ((XElement)node).Name != "a" ? XNodeFilter.SelectNode : XNodeFilter.SkipNode), data.Title);
data.Description = textValues.description;
data.Language = textValues.language;
data.Size = textValues.size;
data.NbPages = textValues.nbPages;
data.Infos.SetValues(textValues.infos);
data.DownloadLinks = xe.XPathValues(".//a/@href").ToArray();
if (__trace)
{
pb.Trace.WriteLine(data.zToJson());
}
return(data);
}
protected override void LoadImages(IPost post)
{
//post.SetImages(DownloadPrint.LoadImages(post.GetImages()));
DownloadPrint.LoadImages(post);
}
protected override RapideDdl_PostDetail GetDataFromWeb(LoadDataFromWeb_v3 loadDataFromWeb)
{
XXElement xeSource = new XXElement(loadDataFromWeb.GetXmlDocument().Root);
RapideDdl_PostDetail data = new RapideDdl_PostDetail();
data.sourceUrl = loadDataFromWeb.request.Url;
data.loadFromWebDate = loadDataFromWeb.loadFromWebDate;
data.id = GetPostDetailKey(data.sourceUrl);
XXElement xePost = xeSource.XPathElement("//div[@class='lcolomn mainside']");
//data.category = xePost.DescendantTextList(".//div[@class='spbar']//a").Select(DownloadPrint.TrimFunc1).Where(
data.category = xePost.XPathElements(".//div[@class='spbar']//a").DescendantTexts().Select(DownloadPrint.Trim).Where(
s =>
{
s = s.ToLowerInvariant();
return(s != "" && !s.Contains("acceuil") && !s.Contains("accueil"));
}
).zToStringValues("/");
string category = data.category.ToLowerInvariant();
data.printType = GetPostType(category);
//data.title = xePost.DescendantTextList(".//div[@class='spbar']", func: DownloadPrint.TrimFunc1).LastOrDefault();
data.title = xePost.XPathElements(".//div[@class='spbar']").DescendantTexts().Select(DownloadPrint.Trim).LastOrDefault();
//ExtractTitleInfos(data);
PrintTitleInfos titleInfos = DownloadPrint.PrintTextValuesManager.ExtractTitleInfos(data.title);
if (titleInfos.foundInfo)
{
data.originalTitle = data.title;
data.title = titleInfos.title;
data.infos.SetValues(titleInfos.infos);
}
XXElement xe = xePost.XPathElement(".//div[@class='shdinfo']");
string date = xe.XPathValue(".//span[@class='date']//text()");
//data.creationDate = Download.Print.RapideDdl.RapideDdl.ParseDateTime(date, loadDataFromWeb.loadFromWebDate);
data.creationDate = zdate.ParseDateTimeLikeToday(date, loadDataFromWeb.loadFromWebDate, "d-M-yyyy, HH:mm", "d M yyyy", "d MMMM yyyy");
if (data.creationDate == null)
{
pb.Trace.WriteLine("unknow date time \"{0}\"", date);
}
if (__trace)
{
pb.Trace.WriteLine("creationDate {0} - \"{1}\"", data.creationDate, date);
}
data.postAuthor = xe.XPathValue(".//span[@class='arg']//a//text()");
xe = xePost.XPathElement(".//div[@class='maincont']");
//data.images = xe.XPathImages(xeImg => new UrlImage(zurl.GetUrl(loadDataFromWeb.request.Url, xeImg.zAttribValue("src")))).ToArray();
data.images = xe.DescendantNodes(node => XmlDescendant.ImageFilter(node)).Select(xeImg => new WebImage(zurl.GetUrl(loadDataFromWeb.request.Url, xeImg.zAttribValue("src")))).ToArray();
if (loadDataFromWeb.request.LoadImage)
{
data.images = DownloadPrint.LoadImages(data.images).ToArray();
}
//RapideDdl.SetTextValues(data, xe.DescendantTextList(nodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a" ));
// xe.DescendantTextList(nodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a")
PrintTextValues_v1 textValues = DownloadPrint.PrintTextValuesManager.GetTextValues_v1(xe.DescendantTexts(node => !(node is XElement) || ((XElement)node).Name != "a" ? XNodeFilter.SelectNode : XNodeFilter.SkipNode), data.title);
data.description = textValues.description;
data.language = textValues.language;
data.size = textValues.size;
data.nbPages = textValues.nbPages;
data.infos.SetValues(textValues.infos);
List <string> downloadLinks = new List <string>();
foreach (XXElement xe2 in xe.XPathElements("div/div"))
{
// http://prezup.eu http://pixhst.com/avaxhome/27/36/002e3627.jpeg http://www.zupmage.eu/i/R1UgqdXn4F.jpg
// http://i.imgur.com/Gu7hagN.jpg http://img11.hostingpics.net/pics/591623liens.png http://www.hapshack.com/images/jUfTZ.gif
// http://pixhst.com/pictures/3029467
downloadLinks.AddRange(xe2.XPathValues(".//a/@href").Where(url => !url.StartsWith("http://prezup.eu") && !url.StartsWith("http://pixhst.com") &&
!url.EndsWith(".jpg") && !url.EndsWith("jpeg") && !url.EndsWith("png") && !url.EndsWith("gif")));
}
data.downloadLinks = downloadLinks.ToArray();
//if (__trace)
// RapideDdl_LoadPostDetail.Trace_RapideDdl_PostDetail(data);
return(data);
}
protected override void LoadImages(RapideDdl_PostDetail post)
{
post.images = DownloadPrint.LoadImages(post.images).ToArray();
}
public static TelechargerMagazine_PostDetail_v1 GetData(WebResult webResult)
{
XXElement xeSource = webResult.Http.zGetXDocument().zXXElement();
TelechargerMagazine_PostDetail_v1 data = new TelechargerMagazine_PostDetail_v1();
data.SourceUrl = webResult.WebRequest.HttpRequest.Url;
data.LoadFromWebDate = webResult.LoadFromWebDate;
data.Id = GetPostDetailKey(webResult.WebRequest.HttpRequest);
// la date est juste la date du jour
// <div id="calendar-layer">
// <table id="calendar" cellpadding="3" class="calendar">
// ...
// <tr>
// ...
// <td class="day-active-v day-current" ><a class="day-active-v" href="http://www.telecharger-magazine.com/2015/07/17/" title="Article posté dans 17 Juillet 2015">17</a></td>
// ...
// </tr>
// ...
// </table>
// </div>
// <div id='dle-content'>
// ...
// <div class="right-full">
//
// <div class="cat_name">
// Posted in:
// <a href="http://www.telecharger-magazine.com/journaux/">Journaux</a>
// </div>
//
// <h2 class="title">
// <img src="/templates/MStarter/images/title.png" alt="" class="img" />
// Journaux Français Du 17 Juillet 2015
// </h2>
//
// <div class="contenttext">
// la date est juste la date du jour
// http://www.telecharger-magazine.com/2015/07/17/
//xeSource.XPathValue("//div[@id='calendar-layer']//table[@id='calendar']//td[@class='day-active-v day-current']//a/@href");
XXElement xePost = xeSource.XPathElement("//div[@id='dle-content']//div[@class='right-full']");
// Journaux
data.Category = xePost.XPathValues(".//div[@class='cat_name']//a/text()").Select(DownloadPrint.Trim).zToStringValues("/");
data.PrintType = GetPrintType(data.Category);
//pb.Trace.WriteLine("category \"{0}\" printType {1}", category, data.printType);
data.Title = xePost.XPathValue(".//h2[@class='title']//text()").zFunc(DownloadPrint.ReplaceChars).zFunc(DownloadPrint.Trim);
PrintTitleInfos titleInfos = DownloadPrint.PrintTextValuesManager.ExtractTitleInfos(data.Title);
if (titleInfos.foundInfo)
{
data.OriginalTitle = data.Title;
data.Title = titleInfos.title;
data.Infos.SetValues(titleInfos.infos);
}
XXElement xeContent = xePost.XPathElement(".//div[@class='contenttext']");
data.Images = new WebImage[] { new WebImage(zurl.GetUrl(data.SourceUrl, xeContent.XPathValue(".//img/@src"))) };
// force load image to get image width and height
if (webResult.WebRequest.LoadImage)
{
data.Images = DownloadPrint.LoadImages(data.Images).ToArray();
}
// get infos, description, language, size, nbPages
PrintTextValues textValues = DownloadPrint.PrintTextValuesManager.GetTextValues(
xeContent.DescendantTexts(
node =>
{
if (node is XText)
{
string text = ((XText)node).Value.Trim();
if (text.ToLowerInvariant() == "description")
{
return(XNodeFilter.DontSelectNode);
}
}
if (node is XElement)
{
XElement xe = (XElement)node;
if (xe.Name == "a")
{
return(XNodeFilter.Stop);
}
}
return(XNodeFilter.SelectNode);
}
).Select(DownloadPrint.ReplaceChars).Select(DownloadPrint.TrimWithoutColon), data.Title, extractValuesFromText: false);
data.Description = textValues.description;
data.Infos.SetValues(textValues.infos);
data.DownloadLinks = xeContent.DescendantNodes(
node =>
{
if (!(node is XElement))
{
return(XNodeFilter.DontSelectNode);
}
XElement xe2 = (XElement)node;
if (xe2.Name == "a")
{
return(XNodeFilter.SelectNode);
}
if (xe2.Name != "p")
{
return(XNodeFilter.DontSelectNode);
}
XAttribute xa = xe2.Attribute("class");
if (xa == null)
{
return(XNodeFilter.DontSelectNode);
}
if (xa.Value != "submeta")
{
return(XNodeFilter.DontSelectNode);
}
//return XNodeFilter.SkipNode;
return(XNodeFilter.Stop);
})
.Select(node => ((XElement)node).Attribute("href").Value).ToArray();
data.DownloadLinks = xeContent.XPathValues(".//a/@href").ToArray();
if (__trace)
{
pb.Trace.WriteLine(data.zToJson());
}
return(data);
}
// detail get data
protected override Vosbooks_PostDetail GetDetailData(WebResult webResult)
{
XXElement xeSource = webResult.Http.zGetXDocument().zXXElement();
Vosbooks_PostDetail data = new Vosbooks_PostDetail();
data.SourceUrl = webResult.WebRequest.HttpRequest.Url;
data.LoadFromWebDate = webResult.LoadFromWebDate;
data.Id = _GetDetailKey(webResult.WebRequest.HttpRequest);
// <div id="page">
// <div id="wrapper">
// <table id="layout">
// <tr>...</tr>
// <tr>
// <td class="sidebars">...</td>
// <td>
// <div id="left-col">
// <div id="content-padding">
// <div id="content">
// ...
// <div class="post" id="post-74299" style="margin-top: 0;">
//
// <table id="post-head">
// <tr>
// <td id="head-date">
// <div class="date"><span>jan</span> 29</div>
// </td>
// <td>
// <div class="title">
// <h2><a href="http://www.vosbooks.net/74299-livre/les-imposteurs-francois-cavanna.html" rel="bookmark" title="Les imposteurs – François Cavanna" >Les imposteurs – François Cavanna </a></h2>
// <div class="postdata">
// <span class="category">
// <a href="http://www.vosbooks.net/category/livre/ebooks-epub" rel="category tag">Ebooks en Epub</a>,
// <a href="http://www.vosbooks.net/category/livre" rel="category tag">Livre</a>
// </span>
// </div>
// </div>
// </td>
// </tr>
// </table>
//
// <div class="entry">
// ...
// <p style="text-align: center;">
// <img class="alignnone" src="http://imageshack.com/a/img538/3859/6JXSxu.jpg" alt="Les imposteurs – François Cavanna" title="Les imposteurs – François Cavanna" height="540" width="420" />
// </p>
// </tr>
XXElement xePost = xeSource.XPathElement("//table[@id='layout']//div[@id='content']//div[@class='post']");
XXElement xe = xePost.XPathElement(".//table[@id='post-head']");
//string[] dates = xe.DescendantTextList(".//td[@id='head-date']", func: Vosbooks.TrimFunc1).ToArray();
string[] dates = xe.XPathElement(".//td[@id='head-date']").DescendantTexts().Select(DownloadPrint.Trim).ToArray();
data.PostCreationDate = GetDate(dates, _lastPostDate);
if (data.PostCreationDate != null)
{
_lastPostDate = new Date(data.PostCreationDate.Value);
}
if (__trace)
{
pb.Trace.WriteLine("post creation date {0} - {1}", data.PostCreationDate, dates.zToStringValues());
}
//data.Title = xePost.XPathValue(".//div[@class='title']//a//text()", DownloadPrint.TrimFunc1);
data.Title = xePost.XPathValue(".//div[@class='title']//a//text()").zFunc(DownloadPrint.ReplaceChars).zFunc(DownloadPrint.Trim);
PrintTitleInfos titleInfos = DownloadPrint.PrintTextValuesManager.ExtractTitleInfos(data.Title);
if (titleInfos.foundInfo)
{
data.OriginalTitle = data.Title;
data.Title = titleInfos.title;
data.Infos.SetValues(titleInfos.infos);
}
// Ebooks en Epub / Livre
//data.Category = xePost.DescendantTextList(".//div[@class='postdata']//span[@class='category']//a").Select(DownloadPrint.TrimFunc1).zToStringValues("/");
data.Category = xePost.XPathElements(".//div[@class='postdata']//span[@class='category']//a").DescendantTexts().Select(DownloadPrint.Trim).zToStringValues("/");
data.PrintType = GetPrintType(data.Category);
//pb.Trace.WriteLine("category \"{0}\" printType {1}", category, data.printType);
xe = xePost.XPathElement(".//div[@class='entry']");
data.Images = new WebImage[] { new WebImage(zurl.GetUrl(data.SourceUrl, xe.XPathValue("div[starts-with(@class, 'post-views')]/following-sibling::h3/following-sibling::p/img/@src"))) };
// force load image to get image width and height
if (webResult.WebRequest.LoadImage)
{
data.Images = DownloadPrint.LoadImages(data.Images).ToArray();
}
// get infos, description, language, size, nbPages
// xe.DescendantTextList(".//p")
PrintTextValues textValues = DownloadPrint.PrintTextValuesManager.GetTextValues(
xe.XPathElements(".//p").DescendantTexts(
node =>
{
if (node is XText)
{
string text = ((XText)node).Value.Trim();
//if (text.StartsWith("Lien Direct", StringComparison.InvariantCultureIgnoreCase))
if (text.StartsWith("lien ", StringComparison.InvariantCultureIgnoreCase))
{
return(XNodeFilter.Stop);
}
}
if (node is XElement)
{
XElement xe2 = (XElement)node;
if (xe2.Name == "p" && xe2.zAttribValue("class") == "submeta")
{
return(XNodeFilter.Stop);
}
}
return(XNodeFilter.SelectNode);
}
).Select(DownloadPrint.ReplaceChars).Select(DownloadPrint.TrimWithoutColon), data.Title);
data.Description = textValues.description;
//data.Language = textValues.language;
//data.Size = textValues.size;
//data.NbPages = textValues.nbPages;
data.Infos.SetValues(textValues.infos);
//data.DownloadLinks = xe.DescendantNodes(
// node =>
// {
// if (!(node is XElement))
// return true;
// XElement xe2 = (XElement)node;
// if (xe2.Name != "p")
// return true;
// XAttribute xa = xe2.Attribute("class");
// if (xa == null)
// return true;
// if (xa.Value != "submeta")
// return true;
// return false;
// },
// node => node is XElement && ((XElement)node).Name == "a")
// .Select(node => ((XElement)node).Attribute("href").Value).ToArray();
data.DownloadLinks = xe.DescendantNodes(
node =>
{
if (!(node is XElement))
{
return(XNodeFilter.DontSelectNode);
}
XElement xe2 = (XElement)node;
if (xe2.Name == "a")
{
return(XNodeFilter.SelectNode);
}
if (xe2.Name != "p")
{
return(XNodeFilter.DontSelectNode);
}
XAttribute xa = xe2.Attribute("class");
if (xa == null)
{
return(XNodeFilter.DontSelectNode);
}
if (xa.Value != "submeta")
{
return(XNodeFilter.DontSelectNode);
}
//return XNodeFilter.SkipNode;
return(XNodeFilter.Stop);
})
.Select(node => ((XElement)node).Attribute("href").Value).ToArray();
//// <div id="postlist" class="postlist restrain">
//xe = xePost.XPathElement(".//div[@id='postlist']");
//// Aujourd'hui, 07h32 - Aujourd'hui, 10h51 - Hier, 12h55 - 22/02/2014, 21h09
////string date = xe.DescendantTextList(".//div[@class='posthead']//text()", nodeFilter: node => node.zGetName() != "a").zToStringValues("");
//XXElement xe2 = xe.XPathElement(".//div[@class='posthead']");
//string date = xe2.DescendantTextList(nodeFilter: node => node.zGetName() != "a").zToStringValues("");
//date = date.Replace('\xA0', ' ');
//data.PostCreationDate = zdate.ParseDateTimeLikeToday(date, webResult.LoadFromWebDate, @"d/M/yyyy, HH\hmm", @"d-M-yyyy, HH\hmm");
//if (data.PostCreationDate == null)
// pb.Trace.WriteLine("unknow post creation date \"{0}\"", date);
//data.PostAuthor = xe.XPathValue(".//div[@class='userinfo']//a//text()", DownloadPrint.TrimFunc1);
//// <div class="postbody">
//xe = xePost.XPathElement(".//div[@class='postbody']//div[@class='content']//blockquote/div");
//data.Images = xe.XPathImages(xeImg => new UrlImage(zurl.GetUrl(data.SourceUrl, xeImg.zAttribValue("src")))).ToArray();
//// get infos, description, language, size, nbPages
//PrintTextValues textValues = DownloadPrint.PrintTextValuesManager.GetTextValues(xe.DescendantTextList(nodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a"), data.Title);
//data.Description = textValues.description;
//data.Language = textValues.language;
//data.Size = textValues.size;
//data.NbPages = textValues.nbPages;
//data.Infos.SetValues(textValues.infos);
//data.DownloadLinks = xe.XPathValues(".//a/@href");
if (__trace)
{
pb.Trace.WriteLine(data.zToJson());
}
return(data);
}
//public string GetOriginalTitle()
//{
// return OriginalTitle;
//}
//public string GetPostAuthor()
//{
// return PostAuthor;
//}
//public DateTime? GetPostCreationDate()
//{
// return PostCreationDate;
//}
//public WebImage[] GetImages()
//{
// return Images;
//}
//public void SetImages(WebImage[] images)
//{
// Images = images;
//}
public void LoadImages()
{
Images = DownloadPrint.LoadImages(Images).ToArray();
}
//protected override Telechargementz_PostDetail GetDataFromWeb(LoadDataFromWeb loadDataFromWeb)
protected override IPost GetDataFromWeb(LoadDataFromWeb_v3 loadDataFromWeb)
{
XXElement xeSource = new XXElement(loadDataFromWeb.GetXmlDocument().Root);
Telechargementz_PostDetail data = new Telechargementz_PostDetail();
data.SourceUrl = loadDataFromWeb.request.Url;
data.LoadFromWebDate = loadDataFromWeb.loadFromWebDate;
data.Id = GetPostDetailKey(data.SourceUrl);
XXElement xePost = xeSource.XPathElement("//div[@id='dle-content']");
data.PostAuthor = xePost.XPathValue(".//div[@class='title-info']//a//text()");
// , 26.12.14
string date = xePost.XPathValue(".//div[@class='title-info']//a/following-sibling::text()");
if (date != null)
{
data.PostCreationDate = zdate.ParseDateTimeLikeToday(date.Trim(' ', ','), loadDataFromWeb.loadFromWebDate, "dd.MM.yy");
if (data.PostCreationDate == null)
{
pb.Trace.WriteLine("unknow date time \"{0}\"", date);
}
if (__trace)
{
pb.Trace.WriteLine("creationDate {0} - \"{1}\"", data.PostCreationDate, date);
}
}
else
{
pb.Trace.WriteLine("creationDate not found \"{0}\"", data.SourceUrl);
}
//data.Title = xePost.XPathElement(".//div[@class='post-title']").DescendantTextList(func: DownloadPrint.TrimFunc1).FirstOrDefault();
data.Title = xePost.XPathElement(".//div[@class='post-title']").DescendantTexts().Select(DownloadPrint.Trim).FirstOrDefault();
PrintTitleInfos titleInfos = DownloadPrint.PrintTextValuesManager.ExtractTitleInfos(data.Title);
if (titleInfos.foundInfo)
{
data.OriginalTitle = data.Title;
data.Title = titleInfos.title;
data.infos.SetValues(titleInfos.infos);
}
XXElement xe = xePost.XPathElement(".//div[starts-with(@id, 'news-id-')]");
if (xe.XElement == null)
{
pb.Trace.WriteLine("element not found \".//div[starts-with(@id, 'news-id-')]\"");
}
//data.Images = new List<UrlImage>();
//data.Images.Add(xe.XPathImages(xeImg => new UrlImage(zurl.GetUrl(loadDataFromWeb.request.Url, xeImg.zAttribValue("src")))).FirstOrDefault());
//data.Images = new UrlImage[] { xe.XPathImages(xeImg => new UrlImage(zurl.GetUrl(loadDataFromWeb.request.Url, xeImg.zAttribValue("src")))).FirstOrDefault() };
WebImage image = xe.DescendantNodes(node => XmlDescendant.ImageFilter(node)).Select(xeImg => new WebImage(zurl.GetUrl(loadDataFromWeb.request.Url, xeImg.zAttribValue("src")))).FirstOrDefault();
if (image != null)
{
data.Images = new WebImage[] { image }
}
;
// force load image to get image width and height
if (loadDataFromWeb.request.LoadImage)
{
data.Images = DownloadPrint.LoadImages(data.Images).ToArray();
}
data.DownloadLinks = xe.XPathValues(".//a/@href").ToArray();
//data.category = xePost.DescendantTextList(".//div[@class='hdiin']//a").Select(DownloadPrint.TrimFunc1).zToStringValues("/");
//string category = data.category.ToLowerInvariant();
//data.printType = GetPrintType(category);
////pb.Trace.WriteLine("category \"{0}\" printType {1}", category, data.printType);
// get infos, description, language, size, nbPages
// nodeFilter: not <a> and not <span>
// nodeFilter: node => !(node is XElement) || (((XElement)node).Name != "a" && ((XElement)node).Name != "span")
// nodeFilter: not <a>
//PrintTextValues_old textValues = DownloadPrint.PrintTextValuesManager.GetTextValues_old(xe.DescendantTextList(nodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a"), data.Title);
PrintTextValues_v1 textValues = DownloadPrint.PrintTextValuesManager.GetTextValues_v1(xe.DescendantTexts(node => !(node is XElement) || ((XElement)node).Name != "a" ? XNodeFilter.SelectNode : XNodeFilter.SkipNode), data.Title);
data.description = textValues.description;
data.language = textValues.language;
data.size = textValues.size;
data.nbPages = textValues.nbPages;
data.infos.SetValues(textValues.infos);
data.PrintType = PrintType.UnknowEBook;
if (data.infos.ContainsKey("Bd") || data.infos.ContainsKey("bd") || data.infos.ContainsKey("BD"))
{
data.PrintType = PrintType.Comics;
}
// Editeur : Presse fr
else if (data.infos.ContainsKey("editeur") && data.infos["editeur"] is ZString && ((string)data.infos["editeur"]).ToLowerInvariant() == "presse fr")
{
data.PrintType = PrintType.Print;
}
else if (data.infos.ContainsKey("isbn"))
{
data.PrintType = PrintType.Book;
}
//pb.Trace.WriteLine(xe.DescendantNodes(returnNodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a").Select(node => new { type = node.NodeType, name = node is XElement ? ((XElement)node).Name.LocalName : null, value = node is XText ? ((XText)node).Value : null }).zToJson());
//pb.Trace.WriteLine(xe.DescendantNodes(nodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a").Select(node => new { type = node.NodeType, name = node is XElement ? ((XElement)node).Name.LocalName : null, value = node is XText ? ((XText)node).Value : null }).zToJson());
//pb.Trace.WriteLine(xe.DescendantNodes(returnNodeFilter: node => node is XText).Select(node => new { type = node.NodeType, name = node is XElement ? ((XElement)node).Name.LocalName : null, value = node is XText ? ((XText)node).Value : null }).zToJson());
//pb.Trace.WriteLine(xe.DescendantNodes(nodeFilter: node => !(node is XElement) || (((XElement)node).Name != "a" && ((XElement)node).Name != "span"), returnNodeFilter: node => node is XText).Select(node => new { type = node.NodeType, name = node is XElement ? ((XElement)node).Name.LocalName : null, value = node is XText ? ((XText)node).Value : null }).zToJson());
if (__trace)
{
pb.Trace.WriteLine(data.zToJson());
}
return(data);
}
public static void LoadImages(IPost post)
{
post.SetImages(DownloadPrint.LoadImages(post.GetImages()).ToArray());
}
protected override void LoadImages(IPost post)
{
//DownloadPrint.LoadImages(post.images);
//post.SetImages(DownloadPrint.LoadImages_new(post.GetImages()).ToArray());
DownloadPrint.LoadImages(post);
}
//protected override GoldenDdl_PostDetail GetDataFromWeb(LoadDataFromWeb loadDataFromWeb)
protected override IPost GetDataFromWeb(LoadDataFromWeb_v3 loadDataFromWeb)
{
XXElement xeSource = new XXElement(loadDataFromWeb.GetXmlDocument().Root);
GoldenDdl_PostDetail data = new GoldenDdl_PostDetail();
data.sourceUrl = loadDataFromWeb.request.Url;
data.loadFromWebDate = loadDataFromWeb.loadFromWebDate;
data.id = GetPostDetailKey(data.sourceUrl);
XXElement xePost = xeSource.XPathElement("//div[@id='dle-content']");
//data.category = xePost.DescendantTextList(".//div[@class='hdiin']//a").Select(DownloadPrint.TrimFunc1).zToStringValues("/");
data.category = xePost.XPathElements(".//div[@class='hdiin']//a").DescendantTexts().Select(DownloadPrint.Trim).zToStringValues("/");
string category = data.category.ToLowerInvariant();
data.printType = GetPrintType(category);
//pb.Trace.WriteLine("category \"{0}\" printType {1}", category, data.printType);
//data.title = xePost.XPathValue(".//div[@class='bheading']//text()", DownloadPrint.Trim);
data.title = xePost.XPathValue(".//div[@class='bheading']//text()").Trim(DownloadPrint.TrimChars);
PrintTitleInfos titleInfos = DownloadPrint.PrintTextValuesManager.ExtractTitleInfos(data.title);
if (titleInfos.foundInfo)
{
data.originalTitle = data.title;
data.title = titleInfos.title;
data.infos.SetValues(titleInfos.infos);
}
string date = xePost.XPathValue(".//div[@class='datenews']//text()");
data.creationDate = zdate.ParseDateTimeLikeToday(date, loadDataFromWeb.loadFromWebDate, "d-M-yyyy, HH:mm", "d M yyyy", "d MMMM yyyy");
if (data.creationDate == null)
{
pb.Trace.WriteLine("unknow date time \"{0}\"", date);
}
if (__trace)
{
pb.Trace.WriteLine("creationDate {0} - \"{1}\"", data.creationDate, date);
}
data.postAuthor = xePost.XPathValue(".//div[@class='argr']//a//text()");
XXElement xe = xePost.XPathElement(".//div[@class='maincont']");
//data.images = xe.XPathImages(xeImg => new UrlImage(zurl.GetUrl(loadDataFromWeb.request.Url, xeImg.zAttribValue("src")))).ToArray();
data.images = xe.DescendantNodes(node => XmlDescendant.ImageFilter(node)).Select(xeImg => new WebImage(zurl.GetUrl(loadDataFromWeb.request.Url, xeImg.zAttribValue("src")))).ToArray();
// force load image to get image width and height
if (loadDataFromWeb.request.LoadImage)
{
data.images = DownloadPrint.LoadImages(data.images).ToArray();
}
// get infos, description, language, size, nbPages
//PrintTextValues_old textValues = DownloadPrint.PrintTextValuesManager.GetTextValues_old(xe.DescendantTextList(nodeFilter: node => !(node is XElement) || ((XElement)node).Name != "a"), data.title);
PrintTextValues_v1 textValues = DownloadPrint.PrintTextValuesManager.GetTextValues_v1(xe.DescendantTexts(node => !(node is XElement) || ((XElement)node).Name != "a" ? XNodeFilter.SelectNode : XNodeFilter.SkipNode), data.title);
data.description = textValues.description;
data.language = textValues.language;
data.size = textValues.size;
data.nbPages = textValues.nbPages;
data.infos.SetValues(textValues.infos);
List <string> downloadLinks = new List <string>();
foreach (XXElement xe2 in xe.XPathElements("div/div"))
{
// http://prezup.eu http://pixhst.com/avaxhome/27/36/002e3627.jpeg http://www.zupmage.eu/i/R1UgqdXn4F.jpg
// http://i.imgur.com/Gu7hagN.jpg http://img11.hostingpics.net/pics/591623liens.png http://www.hapshack.com/images/jUfTZ.gif
// http://pixhst.com/pictures/3029467
downloadLinks.AddRange(xe2.XPathValues(".//a/@href").Where(url => !url.StartsWith("http://prezup.eu") && !url.StartsWith("http://pixhst.com") &&
!url.EndsWith(".jpg") && !url.EndsWith("jpeg") && !url.EndsWith("png") && !url.EndsWith("gif")));
}
data.downloadLinks = downloadLinks.ToArray();
if (__trace)
{
pb.Trace.WriteLine(data.zToJson());
}
return(data);
}
