/// <summary>
/// Izvršava imbBasic metod detekcije templejta
/// </summary>
/// <param name="source">Lista učitanih stranica</param>
/// <param name="settings">Podešavanja</param>
/// <returns>Sređena definicija templejta</returns>
public static templateDefinition detectTemplate_imbBasic(crawledPage[] source, imbWebTemplateSettings settings)
{
templateDefinition output = new templateDefinition();
List <string> xPathList;
// COMMON TREE DETECTION
switch (settings.commonTreeDetection)
{
default:
case commonTreeMethod.imbEndNodePathFrequency:
output.xPathStruktura = templateOperations.commonTree_imbENPF(source, settings);
break;
}
crawledPage c = source.First();
//imbNamespaceSetup nsSetup = new imbNamespaceSetup(c.xmlDocument);
// COMMON CONTENT CHECK
output.xPathStruktura = templateOperations.commonContentCheck(source, output.xPathStruktura, settings);
templateExtensions.prepareContent(output, settings);
// page track
if (settings.doSavePageUrls)
{
foreach (crawledPage p in source)
{
output.relatedPages.Add(p.url);
}
}
output.score = source.Length;
if (output.xPathStruktura.Count == 0)
{
logSystem.log("Template detection failed!", logType.Notification);
}
return(output);
}
/// <summary>
/// Deo imbBasic metodologije
/// </summary>
/// <param name="template">Definicija templejta ciji sadržaj treba podesiti</param>
/// <param name="settings">Podešavanja</param>
public static void prepareContent(templateDefinition template, imbWebTemplateSettings settings)
{
/*
* StringBuilder textMaker = new StringBuilder();
* template.templateXML = new XmlDocument(nsSetup.namespaceManager.NameTable);
*
* String tmp = template.templateXML.OuterXml;
*
* template.templateXML.Prefix = nsSetup.nsPrefix;
*
*
*
* String basicXML = "<?xml version=\"1.0\" encoding=\"utf-16\"?>" + Environment.NewLine;
* basicXML += "<span>" + Environment.NewLine;
* basicXML += "<html xmlns=\"" + nsSetup.nsSourceUrl + "\" >" + Environment.NewLine;
* basicXML += "</html>" + Environment.NewLine;
* basicXML += "</span>" + Environment.NewLine;
*
* template.templateXML.LoadXml(basicXML);
*
*
*
*
* foreach (KeyValuePair<String, templateElement> el in template.xPathStruktura)
* {
* if (settings.doMakeTEXT)
* {
*
* textMaker.AppendLine(el.Value.content);
* }
* if (settings.doMakeXML)
* {
* // imbXmlExtendedTools.makeNodeByxPath(template.templateXML, template.templateXML.DocumentElement, el.Key, el.Value.source, nsSetup);
* }
* }
*
* if (settings.doMakeHTML)
* {
* template.templateHTML = template.templateXML.OuterXml;
* }
* */
}
/// <summary>
/// Description of $property$
/// </summary>
public static string makeUniKey(crawledPage[] input, imbWebTemplateSettings settings,
templateDefinition template)
{
string output = ""; // = new String();
//imbKeywordScoreList nameList = new imbKeywordScoreList();
//switch (settings.uniKeyMode)
//{
// case uniKeyCreation.idToString:
// output = template.id.ToString("D3");
// break;
// case uniKeyCreation.tokenizePageTitles:
// foreach (crawledPage p in input)
// {
// nameList.addText(p.pageCaption, false, imbNLPengine.imbBasic);
// }
// break;
// case uniKeyCreation.tokenizeTemplateContent:
// foreach (crawledPage p in input)
// {
// nameList.addText(template.templateHTML, false, imbNLPengine.imbBasic);
// }
// break;
//}
//if (output == "")
//{
// nameList.sort();
// output = imbCollectionHelpers.imbGetFirstValue<String>(nameList.getStringList(), "", false, 0);
// output = output.TrimToMaxLength(10, "");
//}
//logSystem.log("UniKey created (" + settings.uniKeyMode + ") = " + output, logType.Execution);
return(output);
}
