internal static SortResults SortFixedWidthByKeyDefCore(string sourcefilePath,
SortDefinitions sortDefinitions,
Action<string, string[]> setKeys,
Func<string, bool> dataFilter = null,
string destinationFolder = null,
bool hasHeader = true,
bool returnDuplicates = false,
Action<SortProgress> progress = null,
DataTransportation dataTransportation = null,
bool deleteDbConnPath = true,
bool writeOutSortFile = true,
int maxBatchSize = 250000)
{
ArgumentValidation.Validate(sourcefilePath, setKeys, destinationFolder);
SortVars srtVars = new SortVars(sourcefilePath, destinationFolder);
SortResults srtResults = new SortResults(sourcefilePath, srtVars.DestFolder, srtVars.DbConnPath);
SortProgress srtProgress = new SortProgress();
try
{
srtResults.DeleteDuplicatesFile();
int lineCount = 1;
using (StreamReader reader = new StreamReader(sourcefilePath))
using (SqliteSortDefBulkInserter sortBulkInserter = new SqliteSortDefBulkInserter(srtVars.DbConnPath, sortDefinitions, maxBatchSize))
{
string line;
srtVars.Header = GetHeader(hasHeader, reader);
srtProgress.InitReading();
while ((line = reader.ReadLine()) != null)
{
srtResults.IncrementLinesRead();
ReportReadProgress(progress, srtProgress, srtResults.LinesRead);
if (dataFilter == null || dataFilter(line))
{
string[] keyValues = new string[sortDefinitions.GetKeys().Count];
setKeys(line, keyValues);
sortBulkInserter.Add(new SortKeyData { KeyValues = keyValues, Data = (line + Constants.Common.PreserveCharacter).Compress() });
lineCount++;
}
else
{
srtResults.IncrementFiltered();
}
}
sortBulkInserter.InsertAnyLeftOvers();
sortBulkInserter.AddUnUniqueIndex();
}
srtProgress.InitWriting();
if (writeOutSortFile)
{
srtResults.WriteOutSorted(dbConnPath: srtVars.DbConnPath,
header: srtVars.Header,
sortDefinitions: sortDefinitions,
delimiter: Constants.Delimiters.Tab,
returnDuplicates: returnDuplicates,
dupesFilePath: srtResults.DuplicatesFilePath,
compressed: true,
progress: (counter) => { srtProgress.Counter = counter; if (progress != null) { progress(srtProgress); } },
dataTransportation: dataTransportation,
deleteDb: deleteDbConnPath);
}
else
{
srtResults.Header = srtVars.Header;
}
srtResults.DeleteDuplicatesFileIfNoDuplicates();
}
catch (Exception)
{
CleanUp(srtVars, srtResults);
srtProgress = null;
throw;
}
return srtResults;
}
/// <summary>
/// Sorts a delimited file given a numeric or string key.
/// </summary>
/// <param name="sourcefilePath">Full path and file name of file to be sorted</param>
/// <param name="getKey">Function to construct the key</param>
/// <param name="dataFilter">Function to filter out a data line (true to include data or false to exclude data)</param>
/// <param name="destinationFolder">Folder path where sorted and/or duplicate files will be place. (Uses folder of sourcefilePath when null)</param>
/// <param name="delimiter">Character delimiter</param>
/// <param name="hasHeader">Does the file have a header row</param>
/// <param name="isUniqueKey">If true duplicates will not be included in the sorted file.</param>
/// <param name="returnDuplicates">If true duplicates will be written out to file only if isUniqueKey is true.</param>
/// <param name="sortDir">The sort direction of the key.</param>
/// <param name="progress">A method to report progress</param>
/// <param name="dataTransportation">Define the data transportation method.</param>
/// <param name="maxBatchSize">Control the max insert batch size</param>
internal static SortResults SortDelimitedByKeyCore<T>(
string sourcefilePath,
Func<string[], string, T> getKey,
Func<string[], string, bool> dataFilter = null,
string destinationFolder = null,
string delimiter = Constants.Delimiters.Comma,
bool hasHeader = true,
bool isUniqueKey = false,
bool returnDuplicates = false,
SortDirection sortDir = SortDirection.Ascending,
Action<SortProgress> progress = null,
DataTransportation dataTransportation = null,
bool deleteDbConnPath = true,
bool writeOutSortFile = true,
int maxBatchSize = 250000)
{
ArgumentValidation.Validate<T>(sourcefilePath, getKey, delimiter, destinationFolder, maxBatchSize);
SortVars srtVars = new SortVars(sourcefilePath, destinationFolder);
SortResults srtResults = new SortResults(sourcefilePath, srtVars.DestFolder, srtVars.DbConnPath);
SortProgress srtProgress = new SortProgress();
try
{
srtResults.DeleteDuplicatesFile();
int lineCount = 1;
using (StreamReader reader = new StreamReader(sourcefilePath))
using (SqliteSortKeyBulkInserter<T> sortBulkInserter = new SqliteSortKeyBulkInserter<T>(srtVars.DbConnPath, uniqueKey: isUniqueKey, maxBatchSize: maxBatchSize))
{
string line;
srtVars.Header = GetHeader(hasHeader, reader);
srtProgress.InitReading();
while ((line = reader.ReadLine()) != null)
{
srtResults.IncrementLinesRead();
ReportReadProgress(progress, srtProgress, srtResults.LinesRead);
FileParser.ParseDelimitedString(new StringReader(line), (fields, lNum) =>
{
if (dataFilter == null || dataFilter(fields, line))
{
T key = getKey(fields, line);
sortBulkInserter.Add(key, line + SortFileHelpers.EscapeByDelimiter(delimiter));
lineCount++;
}
else
{
srtResults.IncrementFiltered();
}
}, delimiter);
}
sortBulkInserter.InsertAnyLeftOvers();
sortBulkInserter.AddUnUniqueIndex();
}
srtProgress.InitWriting();
if (writeOutSortFile)
{
srtResults.WriteOutSorted(srtVars.DbConnPath, srtVars.Header, sortDir, delimiter, hasUniqueIndex: isUniqueKey, returnDuplicates: returnDuplicates, dupesFilePath: srtResults.DuplicatesFilePath, progress: (counter) => { srtProgress.Counter = counter; if (progress != null) { progress(srtProgress); } }, dataTransportation: dataTransportation, deleteDb: deleteDbConnPath);
}
else
{
srtResults.Header = srtVars.Header;
}
srtResults.DeleteDuplicatesFileIfNoDuplicates();
}
catch (Exception)
{
CleanUp(srtVars, srtResults);
srtProgress = null;
throw;
}
return srtResults;
}
