private void ReadAndStandarizeIDFile()
{
var distinctGroupQuery = (from q in rawDataFileList where q.selected == true select q.@group).Distinct();
var distinctHashGroupQuery = (from q in rawDataFileList where q.selected == true select q.hashGroup).Distinct();
int counter = 0;
foreach (string currentDistinctGroup in distinctGroupQuery)
{
foreach (string currentDistinctHashGroup in distinctHashGroupQuery)
{
var filesInThisGroup = from q in rawDataFileList
where (q.@group == currentDistinctGroup
&& q.hashGroup == currentDistinctHashGroup
&& q.selected == true)
select q;
DistinctList<StandarizedIDData>[] standarizedIDDataTable = new DistinctList<StandarizedIDData>[hashTableElementCount];
for (int i = 0; i < hashTableElementCount; i++) standarizedIDDataTable[i] = new DistinctList<StandarizedIDData>();
foreach (File currentFile in filesInThisGroup)
{
List<StringDataFormat> stringDataFormatsForCurrentFile = getStringDataFormatsWithRightYear(currentFile);
List<NumberDataFormat> numberDataFormatsForCurrentFile = getNumberDataFormatsWithRightYear(currentFile);
int indexID = stringDataFormatsForCurrentFile.FindIndex(x => x.key == "ID");
int indexBirthday = stringDataFormatsForCurrentFile.FindIndex(x => x.key == "ID_BIRTHDAY");
int indexSex = stringDataFormatsForCurrentFile.FindIndex(x => x.key == "ID_SEX");
int indexInDate = stringDataFormatsForCurrentFile.FindIndex(x => x.key == "ID_IN_DATE");
int indexOutDate = stringDataFormatsForCurrentFile.FindIndex(x => x.key == "ID_OUT_DATE");
using (var sr = new StreamReader(currentFile.path, Encoding.Default))
{
while (!sr.EndOfStream)
{
var dataRow = ReadRow(sr, stringDataFormatsForCurrentFile, numberDataFormatsForCurrentFile);
if (!isMatchBirthYearRange(dataRow.stringData[indexBirthday]))
continue;
var newIDData = new StandarizedIDData()
{
ID = dataRow.stringData[indexID],
Birthday = dataRow.stringData[indexBirthday],
isMale = dataRow.stringData[indexSex] == "M",
firstInDate = dataRow.stringData[indexInDate].StringToDate(),
lastInDate = dataRow.stringData[indexInDate].StringToDate(),
firstOutDate = dataRow.stringData[indexOutDate].StringToDate(),
lastOutDate = dataRow.stringData[indexOutDate].StringToDate()
};
uint hash = getIDHash(newIDData);
int index = standarizedIDDataTable[hash].AddDistinct(newIDData);
if (index >= 0)
{
standarizedIDDataTable[hash][index].isMale = newIDData.isMale;
standarizedIDDataTable[hash][index].firstInDate = newIDData.firstInDate;
standarizedIDDataTable[hash][index].firstOutDate = newIDData.firstOutDate;
standarizedIDDataTable[hash][index].lastInDate = newIDData.lastInDate;
standarizedIDDataTable[hash][index].lastOutDate = newIDData.lastOutDate;
}
else
{
counter++;
}
}
}
}
var outputFilePath = getOutputFilePath(currentDistinctGroup, currentDistinctHashGroup);
using (var sw = new StreamWriter(outputFilePath, false, System.Text.Encoding.Default))
{
sw.WriteLine(StandarizedIDData.ToTitle());
foreach (var thisTable in standarizedIDDataTable)
foreach (var thisstandarizedIDData in thisTable)
{
sw.WriteLine(thisstandarizedIDData.ToWriteLine());
}
}
}
}
}
private uint getIDHash(StandarizedIDData inputIDData) //使用第二組四位數來作為hash (第一組留給split group by hash用)
{
return uint.Parse(inputIDData.ID.Substring(4, 4), System.Globalization.NumberStyles.HexNumber);
}
