public static void Test_Unea_request_01()
{
// req region :
// method post
// url http://www.unea.fr/union-nationale-entreprises-adaptees/annuaire-unea/71/71/annuaire/annuaire.asp
// Cookie __atuvc=1%7C4; PHPSESSID=t85siq9oqn4sqmi0eg09schbh0; ASPSESSIONIDQABATCST=BDBNKJFAHJNFHLOCOJGCGOOF; __utma=169855717.692259951.1390462116.1391519633.1391522222.4; __utmc=169855717; __utmz=169855717.1390462116.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)
// Referer http://www.unea.fr/union-nationale-entreprises-adaptees/annuaire-unea/71/71/annuaire/annuaire.asp?txtRecherche=Entrez%20le%20nom%20d'une%20entreprise&hiddenValider=true
// postData hiddenValider=true&txtRecherche=Par+Nom+Entreprise&txtRecherche1=&txtRecherche2=1&txtRecherche3=&txtRecherche4=
// txtRecherche2=1 ==> 1 = Alsace
string url = "http://www.unea.fr/union-nationale-entreprises-adaptees/annuaire-unea/71/71/annuaire/annuaire.asp";
string request = "hiddenValider=true&txtRecherche=Par+Nom+Entreprise&txtRecherche1=&txtRecherche2=1&txtRecherche3=&txtRecherche4=";
string referer = "http://www.unea.fr/union-nationale-entreprises-adaptees/annuaire-unea/71/71/annuaire/annuaire.asp?txtRecherche=Entrez%20le%20nom%20d'une%20entreprise&hiddenValider=true";
//HtmlXmlReader hr = Http2.HtmlReader;
//hr.WebRequestMethod = HttpRequestMethod.Post;
//hr.WebRequestContent = request;
//hr.WebRequestContentType = "application/x-www-form-urlencoded";
//hr.WebRequestReferer = referer;
HttpRequestParameters_v1 requestParameters = new HttpRequestParameters_v1();
requestParameters.method = HttpRequestMethod.Post;
requestParameters.content = request;
requestParameters.contentType = "application/x-www-form-urlencoded";
requestParameters.referer = referer;
Http_v3.LoadUrl(url, requestParameters);
}
public virtual string GetUrlSubDirectory(string url, HttpRequestParameters_v1 requestParameters = null)
{
if (_getUrlSubDirectory != null)
return _getUrlSubDirectory(url, requestParameters);
else
return null;
}
public override Image LoadImage(string url, HttpRequestParameters_v1 requestParameters = null)
{
MongoImage mongoImage = LoadMongoImage(url, requestParameters);
if (mongoImage.Image == null)
LoadImage(mongoImage);
return mongoImage.Image;
}
//public static CookieContainer Cookies { get { return _hxr.Cookies; } }
public static bool LoadUrl(string url, HttpRequestParameters_v1 requestParameters = null)
{
try
{
_hxr.Load(url, requestParameters);
//WriteLine("request headers :");
//WriteHeaders(_hxr.http.Request.Headers);
//WriteLine("response headers :");
//WriteHeaders(_hxr.http.Response.Headers);
return true;
}
catch (Exception ex)
{
//Load("http://www.frboard.com/magazines-et-journaux/441873-multi-les-journaux-mardi-13-aout-2013-pdf-lien-direct.html");
//15/08/2013 12:00:32 Error : A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond 5.199.168.178:80 (System.Net.Sockets.SocketException)
//Unable to connect to the remote server (System.Net.WebException)
//----------------------
// at System.Net.Sockets.Socket.DoConnect(EndPoint endPointSnapshot, SocketAddress socketAddress)
// at System.Net.ServicePoint.ConnectSocketInternal(Boolean connectFailure, Socket s4, Socket s6, Socket& socket, IPAddress& address, ConnectSocketState state, IAsyncResult asyncResult, Int32 timeout, Exception& exception)
//----------------------
// at System.Net.HttpWebRequest.GetResponse()
// at pb.old.Http.OpenWebRequest() in c:\pib\dropbox\pbeuz\Dropbox\dev\project\Source\Source_01\Source\PB_Tools\\Http_Html.cs:line 911
// at pb.old.Http.Open() in c:\pib\dropbox\pbeuz\Dropbox\dev\project\Source\Source_01\Source\PB_Tools\\Http_Html.cs:line 780
// at pb.old.Http.Load() in c:\pib\dropbox\pbeuz\Dropbox\dev\project\Source\Source_01\Source\PB_Tools\\Http_Html.cs:line 503
// at pb.old.HtmlXmlReader.Load(String sUrl) in c:\pib\dropbox\pbeuz\Dropbox\dev\project\Source\Source_01\Source\PB_Tools\\HtmlXmlReader.cs:line 426
// at Print.download.w.Test_frboard_02()
// at Print.download.w.Run()
//Trace.WriteLine("Error : \"{0}\" ({1})", ex.Message, ex.GetType().ToString());
Trace.CurrentTrace.WriteError(ex);
return false;
}
}
public string GetUrlSubPath(string url, HttpRequestParameters_v1 requestParameters = null)
{
string file = GetUrlFilename(url, requestParameters);
string dir = GetUrlSubDirectory(url, requestParameters);
if (dir != null)
file = zPath.Combine(dir, file);
return file;
}
public MongoImage LoadMongoImage(string url, HttpRequestParameters_v1 requestParameters = null)
{
BsonDocument document = GetCollection().zFindOneById<BsonDocument>(BsonValue.Create(url));
if (document != null)
return BsonSerializer.Deserialize<MongoImage>(document);
else
return CreateMongoImage(url, requestParameters);
}
public static void Test_RapideDdl_LoadUrl_01(string url)
{
//string url = "http://rapide-ddl.com/ebooks/";
HttpRequestParameters_v1 requestParameters = new HttpRequestParameters_v1();
requestParameters.encoding = Encoding.UTF8;
requestParameters.cookies.Add(new Uri(url), new Cookie("hasVisitedSite", "Yes"));
pb.old.Http_v2.LoadUrl(url, requestParameters);
}
public static void Test_UrlToFileName_01()
{
string url = "http://www.unea.fr/union-nationale-entreprises-adaptees/annuaire-unea/71/71/annuaire/annuaire.asp";
string request = "hiddenValider=true&txtRecherche=&txtRecherche1=&txtRecherche2=1&txtRecherche3=&txtRecherche4=";
HttpRequestParameters_v1 requestParameters = new HttpRequestParameters_v1();
requestParameters.method = HttpRequestMethod.Post;
requestParameters.content = request;
//_tr.WriteLine(zurl.UrlToFileName(url, UrlFileNameType.FileName | UrlFileNameType.Content, null, requestParameters));
_tr.WriteLine(zurl.UrlToFileName(new HttpRequest { Url = url, Content = requestParameters.content }, UrlFileNameType.FileName | UrlFileNameType.Content));
}
public static void LoadImages(List<WebImage> images, HttpRequestParameters_v1 requestParameters = null)
{
for (int i = 0; i < images.Count; )
{
if (!LoadImage(images[i], requestParameters, __imageFilter))
images.RemoveAt(i);
else
i++;
}
}
public static HttpRequest CreateHttpRequest(string url, HttpRequestParameters_v1 requestParameters)
{
HttpRequest httpRequest = new HttpRequest();
httpRequest.Url = url;
if (requestParameters != null)
{
httpRequest.Method = requestParameters.method;
httpRequest.Content = requestParameters.content;
httpRequest.Referer = requestParameters.referer;
}
return httpRequest;
}
public static bool LoadToFile(string url, string file, HttpRequestParameters_v1 requestParameters = null)
{
try
{
_hxr.LoadToFile(url, file, requestParameters);
return true;
}
catch (Exception ex)
{
//WriteLine("Error : \"{0}\" ({1})", ex.Message, ex.GetType().ToString());
Trace.CurrentTrace.WriteError(ex);
return false;
}
}
// bool desactivateDocumentStore = false
public static RapideDdl_PostDetail Load(string url, HttpRequestParameters_v1 requestParameters = null, bool reload = false, bool loadImage = false, bool refreshDocumentStore = false)
{
//Trace.WriteLine("RapideDdl_LoadDetail.Load \"{0}\"", url);
if (requestParameters == null)
requestParameters = new HttpRequestParameters_v1();
requestParameters.encoding = Encoding.UTF8;
pb.Web.v1.RequestFromWeb_v2 request = new pb.Web.v1.RequestFromWeb_v2(url, requestParameters, reload, false);
//_load.DesactivateDocumentStore = desactivateDocumentStore;
RapideDdl_PostDetail postDetail = _load.Load(request, GetPostDetailKey(url), refreshDocumentStore);
//_load.DesactivateDocumentStore = false;
if (loadImage)
pb.old.Http_v2.LoadImageFromWeb(postDetail.images);
return postDetail;
}
public static HttpRequestParameters CreateHttpRequestParameters(HttpRequestParameters_v1 requestParameters)
{
HttpRequestParameters requestParameters_new = new HttpRequestParameters();
if (requestParameters != null)
{
requestParameters_new.Accept = requestParameters.accept;
requestParameters_new.AutomaticDecompression = requestParameters.automaticDecompression;
requestParameters_new.ContentType = requestParameters.contentType;
requestParameters_new.Cookies = requestParameters.cookies;
requestParameters_new.Encoding = requestParameters.encoding;
requestParameters_new.Expect100Continue = requestParameters.Expect100Continue;
requestParameters_new.Headers = requestParameters.headers;
requestParameters_new.UserAgent = requestParameters.userAgent;
}
return requestParameters_new;
}
public static TelechargementPlus_PostDetail Load(string url, HttpRequestParameters_v1 requestParameters = null, bool reload = false, bool loadImage = false)
{
//Trace.WriteLine("TelechargementPlus_LoadDetail.Load \"{0}\"", url);
if (requestParameters == null)
requestParameters = new HttpRequestParameters_v1();
requestParameters.encoding = Encoding.UTF8;
//RequestFromWeb request = new RequestFromWeb(url, requestParameters, reload, loadImage);
pb.Web.v1.RequestFromWeb_v2 request = new pb.Web.v1.RequestFromWeb_v2(url, requestParameters, reload, false);
//WebDataRequest<TelechargementPlus_PostDetail> request = new WebDataRequest<TelechargementPlus_PostDetail>(url, GetPostDetailKey(url), requestParameters, reload, loadImage);
TelechargementPlus_PostDetail postDetail = _load.Load(request, GetPostDetailKey(url));
if (loadImage)
{
pb.old.Http_v2.LoadImageFromWeb(postDetail.images);
}
return postDetail;
}
public static bool LoadImage(WebImage image, HttpRequestParameters_v1 requestParameters = null, Predicate<ImageMongoCache_v1> filter = null)
{
if (image.Url != null && image.Image == null)
{
if (__imageCacheManager != null)
{
ImageMongoCache_v1 imageCache = (ImageMongoCache_v1)__imageCacheManager.GetImageCache(image.Url, requestParameters);
if (filter != null && !filter(imageCache))
return false;
image.Image = imageCache.Image;
}
else
image.Image = pb.old.Http_v2.LoadImageFromWeb(image.Url, requestParameters);
}
return true;
}
//public static void LoadImageFromWeb(ImageHtml image, HttpRequestParameters requestParameters = null)
//{
// if (image.Source != null)
// image.Image = LoadImageFromWeb(image.Source, requestParameters);
//}
// List<ImageHtml> images
public static void LoadImageFromWeb(IEnumerable<ImageHtml> images, HttpRequestParameters_v1 requestParameters = null)
{
foreach (ImageHtml image in images)
{
if (image.Source != null)
{
if (image.Image == null)
image.Image = LoadImageFromWeb(image.Source, requestParameters);
if (image.Image != null)
{
image.ImageWidth = image.Image.Width;
image.ImageHeight = image.Image.Height;
}
}
}
}
public static void Test_Handeco_LoadHeaderPages_01(int nbPage = 1, bool loadLastPage = false)
{
Trace.WriteLine("Test_Handeco_Load_01");
// www.handeco.org => 87.98.177.160
// impossible d'aller directement à la page 2
// les paramètres sont envoyés par un post qui renvoie la page 1
// puis on fait un get pour les pages suivantes
string url = "http://www.handeco.org/fournisseurs/rechercher";
//Http2.LoadUrl(url);
// tous les départements
string content = "raisonSociale=&SIRET=&departements%5B%5D=67&departements%5B%5D=68&departements%5B%5D=24&departements%5B%5D=33&departements%5B%5D=40&departements%5B%5D=47&departements%5B%5D=64&departements%5B%5D=03&departements%5B%5D=15&departements%5B%5D=43&departements%5B%5D=63&departements%5B%5D=14&departements%5B%5D=50&departements%5B%5D=61&departements%5B%5D=21&departements%5B%5D=58&departements%5B%5D=71&departements%5B%5D=89&departements%5B%5D=22&departements%5B%5D=29&departements%5B%5D=35&departements%5B%5D=56&departements%5B%5D=18&departements%5B%5D=28&departements%5B%5D=36&departements%5B%5D=37&departements%5B%5D=41&departements%5B%5D=45&departements%5B%5D=08&departements%5B%5D=10&departements%5B%5D=51&departements%5B%5D=52&departements%5B%5D=2A&departements%5B%5D=2B&departements%5B%5D=25&departements%5B%5D=39&departements%5B%5D=70&departements%5B%5D=90&departements%5B%5D=27&departements%5B%5D=76&departements%5B%5D=75&departements%5B%5D=77&departements%5B%5D=78&departements%5B%5D=91&departements%5B%5D=92&departements%5B%5D=93&departements%5B%5D=94&departements%5B%5D=95&departements%5B%5D=11&departements%5B%5D=30&departements%5B%5D=34&departements%5B%5D=48&departements%5B%5D=66&departements%5B%5D=19&departements%5B%5D=23&departements%5B%5D=87&departements%5B%5D=54&departements%5B%5D=55&departements%5B%5D=57&departements%5B%5D=88&departements%5B%5D=09&departements%5B%5D=12&departements%5B%5D=31&departements%5B%5D=32&departements%5B%5D=46&departements%5B%5D=65&departements%5B%5D=81&departements%5B%5D=82&departements%5B%5D=59&departements%5B%5D=62&departements%5B%5D=44&departements%5B%5D=49&departements%5B%5D=53&departements%5B%5D=72&departements%5B%5D=85&departements%5B%5D=02&departements%5B%5D=60&departements%5B%5D=80&departements%5B%5D=16&departements%5B%5D=17&departements%5B%5D=79&departements%5B%5D=86&departements%5B%5D=04&departements%5B%5D=05&departements%5B%5D=06&departements%5B%5D=13&departements%5B%5D=83&departements%5B%5D=84&departements%5B%5D=01&departements%5B%5D=07&departements%5B%5D=26&departements%5B%5D=38&departements%5B%5D=42&departements%5B%5D=69&departements%5B%5D=73&departements%5B%5D=74&departements%5B%5D=971&departements%5B%5D=973&departements%5B%5D=972&departements%5B%5D=974&departements%5B%5D=988&departements%5B%5D=987&departements%5B%5D=975&departements%5B%5D=976&departements%5B%5D=986&experience_cotraitance=0&motsCles=&submitRecherche=Rechercher";
// Auvergne - 03 Allier
//string content = "raisonSociale=&SIRET=&departements%5B%5D=03&experience_cotraitance=0&motsCles=&submitRecherche=Rechercher";
CookieContainer cookies = new CookieContainer();
HttpRequestParameters_v1 requestParameters = new HttpRequestParameters_v1();
requestParameters.cookies = cookies;
requestParameters.method = HttpRequestMethod.Post;
requestParameters.content = content;
//requestParameters.Expect100Continue = true;
//HtmlXmlReader hxr = HtmlXmlReader.CurrentHtmlXmlReader;
//hxr.LoadRepeatIfError = 1;
//hxr.Load(url, requestParameters);
Http_v3.LoadUrl(url, requestParameters);
requestParameters = new HttpRequestParameters_v1();
requestParameters.cookies = cookies;
int page = 1;
for (int i = 1; i < nbPage; i++)
{
// http://www.handeco.org/fournisseurs/rechercher/page/2
url = "http://www.handeco.org/fournisseurs/rechercher/page/" + (++page).ToString();
Http_v3.LoadUrl(url, requestParameters);
}
if (loadLastPage)
{
url = "http://www.handeco.org/fournisseurs/rechercher/page/200";
Http_v3.LoadUrl(url, requestParameters);
}
}
public static void Test_Unea_LoadHeader_01()
{
Unea.Unea.Init();
//string request = "hiddenValider=true&txtRecherche=Par+Nom+Entreprise&txtRecherche1=&txtRecherche2=1&txtRecherche3=&txtRecherche4=";
bool reload = false;
bool loadImage = false;
string url = "http://www.unea.fr/union-nationale-entreprises-adaptees/annuaire-unea/71/71/annuaire/annuaire.asp";
string referer = "http://www.unea.fr/union-nationale-entreprises-adaptees/annuaire-unea/71/71/annuaire/annuaire.asp?txtRecherche=Entrez%20le%20nom%20d'une%20entreprise&hiddenValider=true";
string request = "hiddenValider=true&txtRecherche=&txtRecherche1=&txtRecherche2=1&txtRecherche3=&txtRecherche4=";
string requestContentType = "application/x-www-form-urlencoded";
HttpRequestParameters_v1 requestParameters = new HttpRequestParameters_v1();
requestParameters.method = HttpRequestMethod.Post;
requestParameters.content = request;
requestParameters.contentType = requestContentType;
requestParameters.referer = referer;
Unea_LoadHeader load = new Unea_LoadHeader(url, requestParameters);
load.Load(reload, loadImage);
load.Data.zView();
}
protected MongoImage CreateMongoImage(string url, HttpRequestParameters_v1 requestParameters = null)
{
string file = _urlCache.GetUrlSubPath(url, requestParameters);
string path = zPath.Combine(_urlCache.CacheDirectory, file);
if (!zFile.Exists(path))
//Http2.LoadToFile(url, path, requestParameters);
Http_v3.LoadToFile(url, path, requestParameters);
Image image = null;
if (zFile.Exists(path))
{
try
{
image = zimg.LoadFromFile(path);
}
catch (Exception exception)
{
Trace.WriteLine("error unable to load image url \"{0}\" to \"{1}\" (WebImageMongoCacheManager.CreateMongoImage())", url, path);
Trace.Write("error : ");
Trace.WriteLine(exception.Message);
}
}
else
{
Trace.WriteLine("error unable to load image url \"{0}\" to \"{1}\" (WebImageMongoCacheManager.CreateMongoImage())", url, path);
}
MongoImage mongoImage = new MongoImage();
mongoImage.Url = url;
mongoImage.File = file;
mongoImage.Width = image != null ? image.Width : 0;
mongoImage.Height = image != null ? image.Height : 0;
mongoImage.Image = image;
GetCollection().zUpdate(new QueryDocument { { "_id", BsonValue.Create(url) } }, new UpdateDocument { { "$set", mongoImage.ToBsonDocument() } }, UpdateFlags.Upsert);
return mongoImage;
}
public ImageMongoCache_v1(WebImageCacheManager_v1 imageUrlCacheManager, string url, HttpRequestParameters_v1 requestParameters = null)
: base(imageUrlCacheManager, url, requestParameters)
{
}
public override ImageCache_v1 GetImageCache(string url, HttpRequestParameters_v1 requestParameters = null)
{
return new ImageMongoCache_v1(this, url, requestParameters);
}
public static Image LoadImage(string url, HttpRequestParameters_v1 requestParameters = null)
{
try
{
if (__imageCacheManager != null)
{
ImageCache_v1 imageCache = __imageCacheManager.GetImageCache(url, requestParameters);
return imageCache.Image;
}
else
return pb.old.Http_v2.LoadImageFromWeb(url, requestParameters);
}
catch (Exception ex)
{
pb.Trace.WriteLine("error RapideDdl loading image : {0}", ex.Message);
return null;
}
}
public virtual ImageCache_v1 GetImageCache(string url, HttpRequestParameters_v1 requestParameters = null)
{
return new ImageCache_v1(this, url, requestParameters);
}
public static Image LoadImageFromWeb(string url, HttpRequestParameters_v1 requestParameters = null)
{
try
{
Image image = _hxr.LoadImage(url, requestParameters);
//if (image.Height > 200)
// image = image.zResize(height: 200);
return image;
}
catch (Exception ex)
{
//WriteLine("Error : \"{0}\" ({1})", ex.Message, ex.GetType().ToString());
Trace.CurrentTrace.WriteError(ex);
return null;
}
}
public virtual string GetUrlFilename(string url, HttpRequestParameters_v1 requestParameters = null)
{
return zurl.UrlToFileName(url, _urlFileNameType, httpRequestContent: requestParameters != null ? requestParameters.content : null);
}
public abstract Image LoadImage(string url, HttpRequestParameters_v1 requestParameters = null);
public ImageCache_v1(WebImageCacheManager_v1 imageUrlCacheManager, string url, HttpRequestParameters_v1 requestParameters = null)
{
_imageUrlCacheManager = imageUrlCacheManager;
_url = url;
_requestParameters = requestParameters;
}
public string GetUrlPath(string url, HttpRequestParameters_v1 requestParameters = null)
{
return zPath.Combine(_cacheDirectory, GetUrlSubPath(url, requestParameters));
}
public static void Test_frboard_03()
{
//string url = "http://www.frboard.com/magazines-et-journaux/441873-multi-les-journaux-mardi-13-aout-2013-pdf-lien-direct.html";
//string url = "http://www.frboard.com/magazines-et-journaux/442257-multi-les-journaux-mercredi-14-aout-2013-pdf-lien-direct.html";
//string url = "http://www.frboard.com/magazines-et-journaux/442270-multi-le-monde-du-jeudi-15-aout-2013-pdf.html";
string url = "http://www.frboard.com/magazines-et-journaux/442649-multi-les-magazines-jeudi-15-aout-2013-pdf-liens-direct-new-post.html";
//string url = "";
//string url = "";
//string url = "";
//string url = "";
//HtmlXmlReader.CurrentHtmlXmlReader.Cookies.zAdd("http://www.frboard.com/", Frboard.GetCookies());
HttpRequestParameters_v1 requestParameters = new HttpRequestParameters_v1();
requestParameters.cookies.zAdd("http://www.frboard.com/", Frboard.GetCookies());
pb.old.HtmlXmlReader.CurrentHtmlXmlReader.Load(url, requestParameters);
string xpath = "//div[@class='postbody']//div[@class='postrow has_after_content']";
XDocument doc = pb.old.HtmlXmlReader.CurrentHtmlXmlReader.XDocument;
XElement xe = doc.Root.zXPathElement(xpath);
if (xe == null)
{
Trace.WriteLine("node not found \"{0}\"", xpath);
return;
}
xpath = ".//div[@class='content']";
xe = xe.XPathSelectElement(xpath);
if (xe == null)
{
Trace.WriteLine("node not found \"{0}\"", xpath);
return;
}
//foreach (XXNode child in xe.DescendantNodes().zWhereSelect(FrboardPost.Filter))
foreach (pb.old.XXNode child in FrboardPostFilter.GetFilteredNodeList(xe))
{
Trace.WriteLine(child.ToString());
}
}
//public static void Test_frboard_01()
//{
// string url = "http://www.frboard.com/magazines-et-journaux/";
// HtmlXmlReader.CurrentHtmlXmlReader.Load(url);
// HtmlXmlReader.CurrentHtmlXmlReader.ReadSelect("//li[starts-with(@class, 'threadbit')]",
// ".//h3[@class='threadtitle']//a/@href:.:n(href)",
// ".//h3[@class='threadtitle']//a/text():.:n(label1)",
// ".//div[@class='author']//text():.:Concat( ):n(author)");
// //HtmlXmlReader.CurrentHtmlXmlReader.ReadSelect("//span[@class='prev_next']//a[@rel='next']/@href");
//}
public static void Test_frboard_02()
{
//string url = "http://www.frboard.com/magazines-et-journaux/441873-multi-les-journaux-mardi-13-aout-2013-pdf-lien-direct.html";
//string url = "http://www.frboard.com/magazines-et-journaux/442257-multi-les-journaux-mercredi-14-aout-2013-pdf-lien-direct.html";
//string url = "http://www.frboard.com/magazines-et-journaux/442270-multi-le-monde-du-jeudi-15-aout-2013-pdf.html";
string url = "http://www.frboard.com/magazines-et-journaux/442649-multi-les-magazines-jeudi-15-aout-2013-pdf-liens-direct-new-post.html";
//string url = "";
//string url = "";
//string url = "";
//string url = "";
//HtmlXmlReader.CurrentHtmlXmlReader.Cookies.zAdd("http://www.frboard.com/", Frboard.GetCookies());
HttpRequestParameters_v1 requestParameters = new HttpRequestParameters_v1();
requestParameters.cookies.zAdd("http://www.frboard.com/", Frboard.GetCookies());
pb.old.HtmlXmlReader.CurrentHtmlXmlReader.Load(url, requestParameters);
// <div class="postbody"> <div class="content"> <blockquote class="postcontent restore ">
// <div class="postbody"> <div class="postrow has_after_content">
//HtmlXmlReader.CurrentHtmlXmlReader.ReadSelect("//div[@class='postbody']//blockquote");
//string xpath = "//div[@class='postbody']//blockquote";
//string xpath = "//div[@class='postbody']//div[@class='postrow has_after_content']";
string xpath = "//div[@class='postbody']//div[@class='postrow has_after_content']";
XDocument doc = pb.old.HtmlXmlReader.CurrentHtmlXmlReader.XDocument;
XElement xe = doc.Root.zXPathElement(xpath);
if (xe == null)
{
Trace.WriteLine("node not found \"{0}\"", xpath);
return;
}
Trace.WriteLine("found \"{0}\"", xpath);
Trace.WriteLine("date \"{0}\" time \"{1}\"", doc.Root.zXPathValue("//span[@class='date']//text()"), doc.Root.zXPathValue("//span[@class='time']//text()"));
// <div class="userinfo"> <div class="username_container"> <div class="popupmenu memberaction">
// <a rel="nofollow" class="username offline popupctrl" href="http://www.frboard.com/members/145457-lc.good.day.html" title="LC.GooD.Day est déconnecté"><strong>
// <img src="http://www.frboard.com/images/misc/ur/general.png" class="userrank"></img><strong><font color="#696969">LC.GooD.Day</font></strong></strong></a>
Trace.WriteLine("author \"{0}\"", doc.Root.zXPathValue("//div[@class='userinfo']//a//text()"));
xpath = ".//div[@class='content']";
xe = xe.XPathSelectElement(xpath);
if (xe == null)
{
Trace.WriteLine("node not found \"{0}\"", xpath);
return;
}
foreach (XNode child in xe.DescendantNodes())
{
if (child is XElement)
{
XElement xeChild = child as XElement;
//Trace.WriteLine("child element \"{0}\"", xeChild.Name);
if (xeChild.Name == "img")
Trace.WriteLine("img src \"{0}\" alt \"{1}\" title \"{2}\" class \"{3}\"",
xeChild.zAttribValue("src"), xeChild.zAttribValue("alt"), xeChild.zAttribValue("title"), xeChild.zAttribValue("class"));
}
else if (child is XText)
{
XText xtextChild = child as XText;
if (child.Parent.Name == "a")
{
Trace.WriteLine("link \"{0}\" href \"{1}\" rel \"{2}\"", xtextChild.Value, child.Parent.zAttribValue("href"), child.Parent.zAttribValue("rel"));
}
else
Trace.WriteLine("text \"{0}\"", xtextChild.Value.Trim());
}
else
Trace.WriteLine("child \"{0}\"", child.NodeType);
}
Trace.WriteLine();
//HtmlXmlReader.CurrentHtmlXmlReader.ReadSelect("//li[starts-with(@class, 'threadbit')]",
// ".//h3[@class='threadtitle']//a/@href:.:n(href)",
// ".//h3[@class='threadtitle']//a/text():.:n(label1)",
// ".//div[@class='author']//text():.:Concat( ):n(author)");
//HtmlXmlReader.CurrentHtmlXmlReader.ReadSelect("//span[@class='prev_next']//a[@rel='next']/@href");
}
