public fileunit toNewOutput(String content, String name, String extension = "txt")
{
fileunit funit = getNewOutput(name, extension);
funit.setContent(content);
return(funit);
}
private void reportTarget(spiderTarget t, folderNode fn, int c)
{
string pageFolder = "P" + c.ToString("D3") + "_" + t.IsRelevant.ToString();
folderNode pfn = fn.Add(pageFolder, "Page " + c.ToString(), "Report on page " + t.url + " crawled by " + name + ". Target.IsRelevant: " + t.IsRelevant + ".".addLine(pageDescription));
fileunit content = new fileunit(pfn.pathFor("content.txt"), false);
fileunit links = new fileunit(pfn.pathFor("links.txt"), false);
if (t.evaluation != null)
{
t.evaluation.saveObjectToXML(pfn.pathFor("relevance.xml"));
}
content.setContent(t.pageText);
//t.page.relationship.outflowLinks
if (t.page != null)
{
foreach (spiderLink ln in t.page.relationship.outflowLinks.items.Values)
{
string rl = ln.url;
links.Append(ln.url);
}
//t.page.webpage.links.ForEach(x => links.Append(x.nature + " | " + x.name + " | " + x.url));
}
content.Save();
links.Save();
// marks.Save();
}
public void reportCrawler(modelSpiderTestRecord tRecord)
{
folderNode fn = folder[DRFolderEnum.crawler];
string fileprefix = tRecord.instance.name.getCleanFilePath(); //tRecord.name.getCleanFilepath();
if (REPORT_TIMELINE)
{
DataTable timeline = timeSeries.GetAggregatedTable("frontier_stats", dataPointAggregationAspect.overlapMultiTable); //.GetSumTable("timeline_" + fileprefix.Replace(" ", ""));
timeline.GetReportAndSave(folder[DRFolderEnum.crawler], notation, "frontier_stats" + fileprefix);
}
if (REPORT_ITERATION_URLS)
{
tRecord.allUrls = urlsLoaded.GetAllUnique();
tRecord.allDetectedUrls = urlsDetected.GetAllUnique();
saveOutput(tRecord.allDetectedUrls, folder[DRFolderEnum.crawler].pathFor("urls_detected.txt"));
saveOutput(tRecord.allUrls, folder[DRFolderEnum.crawler].pathFor("urls_loaded.txt"));
saveOutput(tRecord.relevantPages, folder[DRFolderEnum.crawler].pathFor("urls_relevant_loaded.txt"));
}
// Int32 iterations = tRecord.instance.settings.limitIterations;
DataTable cpuTable = tRecord.cpuTaker.GetDataTableBase("cpuMetrics").GetReportAndSave(folder[DRFolderEnum.crawler], notation, "cpu_" + fileprefix);
DataTable dataTable = tRecord.dataLoadTaker.GetDataTableBase("dataLoadMetrics").GetReportAndSave(folder[DRFolderEnum.crawler], notation, "dataload_" + fileprefix);
DataTable resourcesTable = tRecord.measureTaker.GetDataTableBase("resourceMetrics").GetReportAndSave(folder[DRFolderEnum.crawler], notation, "resource_" + fileprefix);
if (imbWEMManager.settings.directReportEngine.doPublishPerformance)
{
tRecord.performance.folderName = folder.name;
tRecord.performance.deploy(tRecord);
tRecord.performance.saveObjectToXML(folder[DRFolderEnum.crawler].pathFor("performance.xml"));
DataTable pTable = tRecord.performance.GetDataTable(true).GetReportAndSave(folder, notation, "crawler_performance" + fileprefix);
}
tRecord.lastDomainIterationTable.GetDataTable(null, imbWEMManager.index.experimentEntry.CrawlID).GetReportAndSave(folder, notation, "DLCs_performance_" + fileprefix);
tRecord.reporter = this;
signature.deployReport(tRecord);
//signature.notation = notation;
signature.saveObjectToXML(folder.pathFor("signature.xml"));
folder.generateReadmeFiles(notation);
fileunit tLog = new fileunit(folder[DRFolderEnum.logs].pathFor(fileprefix + ".txt"), false);
tLog.setContent(tRecord.logBuilder.ContentToString(true));
tLog.Save();
tRecord.instance.reportCrawlFinished(this, tRecord);
aceLog.consoleControl.setLogFileWriter();
}
/// <summary>
/// Runs when a DLC is finished
/// </summary>
/// <param name="wRecord">The w record.</param>
public void reportDomainFinished(modelSpiderSiteRecord wRecord)
{
folderNode fn = null;
string fileprefix = wRecord.domainInfo.domainRootName.getCleanFilepath();
if (imbWEMManager.settings.directReportEngine.doDomainReport)
{
fn = folder[DRFolderEnum.sites].Add(wRecord.domainInfo.domainRootName.getCleanFilepath(), "Report on " + wRecord.domainInfo.domainName, "Records on domain " + wRecord.domainInfo.domainName + " crawled by " + name);
if (REPORT_DOMAIN_TERMS)
{
if (wRecord.tRecord.instance.settings.doEnableDLC_TFIDF)
{
if (wRecord.context.targets.dlTargetPageTokens != null)
{
wRecord.context.targets.dlTargetPageTokens.GetDataSet(true).serializeDataSet("token_ptkn", fn, dataTableExportEnum.excel, notation);
}
}
if (wRecord.context.targets.dlTargetLinkTokens != null)
{
wRecord.context.targets.dlTargetLinkTokens.GetDataSet(true).serializeDataSet("token_ltkn", fn, dataTableExportEnum.excel, notation);
}
}
if (REPORT_DOMAIN_PAGES)
{
int c = 1;
foreach (spiderTarget t in wRecord.context.targets.GetLoadedInOrderOfLoad())
{
reportTarget(t, fn, c);
c++;
}
}
fileunit wLog = new fileunit(folder[DRFolderEnum.logs].pathFor(fileprefix + ".txt"), false);
wLog.setContent(wRecord.logBuilder.ContentToString(true));
wLog.Save();
if (REPORT_ITERATION_URLS)
{
textByIteration url_loaded = urlsLoaded[wRecord]; //.GetOrAdd(wRecord, new textByIteration());
textByIteration url_detected = urlsDetected[wRecord]; //, new textByIteration());
fileunit url_ld_out = new fileunit(folder[DRFolderEnum.sites].pathFor(fileprefix + "_url_ld.txt"), false);
fileunit url_dt_out = new fileunit(folder[DRFolderEnum.sites].pathFor(fileprefix + "_url_dt.txt"), false);
fileunit url_srb_out = new fileunit(folder[DRFolderEnum.sites].pathFor(fileprefix + "_url_srb_ld.txt"), false);
url_ld_out.setContentLines(url_loaded.GetAllUnique());
url_dt_out.setContentLines(url_detected.GetAllUnique());
url_srb_out.setContentLines(wRecord.relevantPages);
url_ld_out.Save();
url_dt_out.Save();
url_srb_out.Save();
}
//terms_out.Save();
//sentence_out.Save();
}
if (REPORT_MODULES)
{
if (wRecord.tRecord.instance is spiderModularEvaluatorBase)
{
wRecord.frontierDLC.reportDomainOut(wRecord, fn, fileprefix);
}
}
if (REPORT_TIMELINE)
{
wRecord.iterationTableRecord.GetDataTable(null, "iteration_performace_" + fileprefix).GetReportAndSave(folder[DRFolderEnum.it], notation, "iteration_performace_" + fileprefix); //, notation);
}
//if (REPORT_TIMELINE)
//{
// DataTable dt = wRecord.GetTimeSeriesPerformance();
// timeSeries.Add(dt);
// dt.GetReportAndSave(folder[DRFolderEnum.it], notation, "iteration_frontier_stats_" + fileprefix);
//}
wRecord.tRecord.lastDomainIterationTable.Add(wRecord.iterationTableRecord.GetLastEntryTouched());
wRecord.tRecord.instance.reportDomainFinished(this, wRecord);
wRecord.Dispose();
}