public void RunMainProcess()
{
LanguageDomainReport firstReport = new LanguageDomainReport();
LanguagePageReport secondReport = new LanguagePageReport();
string targetFile;
DateTime endPeriod = StartPeriod.AddDays(this.DaysNumber * Config.OperatorBackDays);
DateTime startPeriod = StartPeriod;
string directoryPath = FileData.GetDataPath(startPeriod);
PageViewsRawData pageViewData = new PageViewsRawData(GZip, FileData, this.DaysNumber);
pageViewData.DownloadAllPeriodsData(startPeriod, endPeriod, directoryPath);
pageViewData.DecompressData(directoryPath);
CollectionsData processData = new CollectionsData(domainData);
startPeriod = StartPeriod;
//Process each file (period) and get the ressults
do
{
FileData.SetConfigurationByPeriod(startPeriod, directoryPath);
targetFile = FileData.GetTargetFileNoExt();
//for free memory related to last analysis
using (FileReader fr = new FileReader(targetFile))
{
Console.WriteLine(string.Concat("Processing data in file ", targetFile, "..."));
//container of data to analyze
PageViewCollection periodCollection = fr.GetDataToCollection();
//First report
Console.WriteLine("\tProcessing language and domain data for period " + startPeriod.ToShortDateString());
LanguageDomain languageDomain = processData.GetLanguageAndDomainCount(periodCollection);
languageDomain.Period = startPeriod;
firstReport.AddLanguageDomain(languageDomain);
//Second report
LanguagePage languagePage = processData.GetLanguagePageCount(periodCollection);
Console.WriteLine("\tProcessing language page data for period " + startPeriod.ToShortDateString());
languagePage.Period = startPeriod;
secondReport.AddLanguagePage(languagePage);
}
startPeriod = startPeriod.AddDays(Config.OperatorBackDays);
}while (startPeriod.CompareTo(endPeriod) != 0);
//Report the results
Console.Clear();
ConsoleReport.DisplayLanguageDomainCount(firstReport);
ConsoleReport.DisplayLanguagePageMaxCount(secondReport);
}
public LanguagePage GetLanguagePageCount(PageViewCollection periodCollection)
{
LanguagePage report = new LanguagePage();
var entry = periodCollection.Data.Where(e => e.ViewCount > 0)
.GroupBy(x => x.PageTitle)
.Select(y => new PageViewEntry
{
PageTitle = y.Key,
ViewCount = Convert.ToUInt32(y.Sum(s => s.ViewCount)),
})
.OrderByDescending(x => x.ViewCount)
.FirstOrDefault();
report.Page = entry.PageTitle; //page tittle
report.ViewCount = entry.ViewCount;
return(report);
}
public PageViewCollection GetDataToCollection()
{
this.fileData = new PageViewCollection();
Console.WriteLine(string.Concat("Reading file ", targetFileNoExt, "..."));
using (FileStream fs = File.Open(targetFileNoExt, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))
using (BufferedStream bs = new BufferedStream(fs))
using (StreamReader sr = new StreamReader(bs))
{
string line;
while ((line = sr.ReadLine()) != null)
{
//TODO: can optimize memory usage, by injecting the logix line by line
fileData.AddPageView(GetEntry(line));
}
}
return(fileData);
}
public LanguageDomain GetLanguageAndDomainCount(PageViewCollection periodCollection)
{
LanguageDomain report = new LanguageDomain();
var entry = periodCollection.Data.Where(e => e.ViewCount > 0)
.GroupBy(x => x.DomainCode)
.Select(y => new PageViewEntry
{
DomainCode = y.Key,
ViewCount = Convert.ToUInt32(y.Sum(s => s.ViewCount)),
})
.OrderByDescending(x => x.ViewCount)
.FirstOrDefault();
report.Domain = domainData.GetDomainNameByCode(entry.DomainCode); //domain
report.LanguageCode = domainData.GetLanguage(entry.DomainCode); //language from domain
report.ViewCount = entry.ViewCount;
return(report);
}