// header get data, from WebHeaderDetailMongoManagerBase_v2<THeaderData, TDetailData>
protected override IEnumDataPages <Handeco_Header_v2> GetHeaderPageData(HttpResult <string> httpResult)
{
XXElement xeSource = httpResult.zGetXDocument().zXXElement();
string url = httpResult.Http.HttpRequest.Url;
Handeco_HeaderDataPages data = new Handeco_HeaderDataPages();
data.SourceUrl = url;
data.LoadFromWebDate = httpResult.Http.RequestTime;
data.Id = GetPageKey(httpResult.Http.HttpRequest);
// <div class="paginationControl">
// page n : <a href="/fournisseurs/rechercher/page/2#resultats">></a> |
// last page : <span class="disabled">></span> |
data.UrlNextPage = zurl.RemoveFragment(zurl.GetUrl(url, xeSource.XPathValue("//div[@class='paginationControl']//*[position()=last()-1]/@href")));
IEnumerable <XXElement> xeHeaders = xeSource.XPathElements("//table//tr[position() > 1]");
List <Handeco_Header_v2> headers = new List <Handeco_Header_v2>();
foreach (XXElement xeHeader in xeHeaders)
{
Handeco_Header_v2 header = new Handeco_Header_v2();
header.SourceUrl = url;
header.LoadFromWebDate = DateTime.Now;
header.Name = Handeco.Trim(xeHeader.XPathValue(".//td[1]//text()"));
header.UrlDetail = zurl.RemoveFragment(zurl.GetUrl(url, xeHeader.XPathValue(".//td[1]//a/@href")));
//header.Siret = Handeco.Trim(xeHeader.XPathValue(".//td[2]//text()"));
header.Type = Handeco.Trim(xeHeader.XPathValue(".//td[2]//text()"));
header.Groupes = xeHeader.XPathValues(".//td[3]//text()").Select(Handeco.Trim).ToArray();
header.Activités = xeHeader.XPathValues(".//td[4]//text()").Select(Handeco.Trim).ToArray();
header.PostalCode = Handeco.Trim(xeHeader.XPathValue(".//td[5]//text()"));
headers.Add(header);
}
data.Data = headers.ToArray();
return(data);
}
private static string[] GetTextValues(XXElement xe)
{
return(xe.XPathValues(".//td//text()").Select(s => Handeco.Trim(_badCharacters.Replace(s, " "))).ToArray());;
}
protected void _GetDetailData(XXElement xeSource, Handeco_Detail_v2 data)
{
//<div style="text-align: right; font-size: 10px;">
//<em>Dernière mise à jour le 18-01-2013</em>
//</div>
string lastUpdate = Handeco.Trim(xeSource.XPathValue("//em[starts-with(text(), 'Dernière mise à jour')]/text()"));
if (lastUpdate != null)
{
Match match = _lastUpdateRegex.Match(lastUpdate);
DateTime date;
if (match.Success && DateTime.TryParseExact(match.Value, "dd-MM-yyyy", System.Globalization.CultureInfo.CurrentCulture, System.Globalization.DateTimeStyles.None, out date))
{
data.DernièreMiseàjour = date;
}
else
{
data.UnknowInfos.Add(lastUpdate);
}
}
else
{
pb.Trace.WriteLine("error \"Dernière mise à jour\" not found");
}
// NOTRE OFFRE - activities - multiple
//<select style="width: 200px; display: none;" onchange="change_activite(this.selectedIndex);" id="select_activites">
// <option>Sous-traitance industrielle - Autre</option>
// <option>Assemblage mécanique</option>
// <option>Energie renouvelable - Autre</option>
//</select>
string[] activityTypes = xeSource.XPathValues("//select[@id = 'select_activites']/option/text()").Select(Handeco.Trim).ToArray();
// CONTACTS - multiple
//<select style="width: 200px; display: none;" onchange="change_contact(this.selectedIndex);" id="select_contacts">
// <option>Jacky STEINLE (Chef d'atelier)</option>
//</select>
string[] contactDescriptions = xeSource.XPathValues("//select[@id = 'select_contacts']/option/text()").Select(Handeco.Trim).ToArray();
int indexActivityType = 0;
int indexContactDescription = 0;
List <Activity> activities = new List <Activity>();
List <Contact> contacts = new List <Contact>();
foreach (XXElement xxe in xeSource.XPathElements("//table[@class = 'fiche organisation']"))
{
//string id = xxe.XPathValue("@id").ToLower();
string id = xxe.XPathValue("@id");
if (id != null)
{
id = id.ToLower();
}
//if (__trace)
// pb.Trace.WriteLine("table id = \"{0}\"", id);
Activity activity = null;
Contact contact = null;
if (id != null && id.StartsWith("fiche_activite_"))
{
activity = new Activity();
activities.Add(activity);
if (indexActivityType < activityTypes.Length)
{
activity.Type = activityTypes[indexActivityType++];
}
else
{
pb.Trace.WriteLine("warning miss an activity type in html (<select id='select_activites'>)");
}
}
else if (id != null && id.StartsWith("fiche_contact_"))
{
contact = new Contact();
contacts.Add(contact);
if (indexContactDescription < contactDescriptions.Length)
{
contact.Description = contactDescriptions[indexContactDescription++];
}
else
{
pb.Trace.WriteLine("warning miss an activity type in html (<select id='select_contacts'>)");
}
}
foreach (XXElement xxe2 in xxe.XPathElements(".//tr"))
{
string valueName = Handeco.Trim(xxe2.XPathValue(".//th//text()"));
//_currentElement = xxe2;
XXElement currentElement = xxe2;
if (valueName == null)
{
continue;
}
if (activity != null)
{
//if (__trace)
// pb.Trace.Write("activité ");
if (!SetActivityValue(activity, valueName, currentElement))
{
//if (__trace)
// pb.Trace.Write("error ");
data.UnknowInfos.Add("valeur activité inconnu : " + valueName + " = " + GetTextValue(currentElement));
}
//else if (__trace)
// pb.Trace.Write(" ");
//if (__trace)
// pb.Trace.WriteLine("\"{0}\" = \"{1}\"", valueName, GetTextValue(currentElement));
}
else if (contact != null)
{
//if (__trace)
// pb.Trace.Write("contact ");
if (!SetContactValue(contact, valueName, currentElement))
{
//if (__trace)
// pb.Trace.Write("error ");
data.UnknowInfos.Add("valeur contact inconnu : " + valueName + " = " + GetTextValue(currentElement));
}
//else if (__trace)
// pb.Trace.Write(" ");
//if (__trace)
// pb.Trace.WriteLine("\"{0}\" = \"{1}\"", valueName, GetTextValue(currentElement));
}
else
{
//if (__trace)
// pb.Trace.Write("société ");
if (!SetValue(data, valueName, currentElement))
{
//if (__trace)
// pb.Trace.Write("error ");
data.UnknowInfos.Add("valeur inconnu : " + valueName + " = " + GetTextValue(currentElement));
}
//else if (__trace)
// pb.Trace.Write(" ");
//if (__trace)
// pb.Trace.WriteLine("\"{0}\" = \"{1}\"", valueName, GetTextValue(currentElement));
}
}
}
data.Activités = activities.ToArray();
data.Contacts = contacts.ToArray();
}