public void ExportFieldsToTsv(string inputFilePath, string outputFilePath, string recordType, List <string> outputColumns)
{
var sections = this.Parser.RetrieveSections(inputFilePath, recordType);
using (var outputFile = new StreamWriter(outputFilePath))
{
foreach (var section in sections)
{
var tsvLine = TsvUtils.GenerateTsvLine(section.Records, outputColumns, removeDoubleQuotes: true);
outputFile.WriteLine(tsvLine);
}
}
}
protected void NetworksLocationPropertyCountsToTsv(WhoisParser parser, string inputFilePath, string propertyName, string outputFilePath)
{
var outputFolderPath = Path.GetDirectoryName(outputFilePath);
if (!Directory.Exists(outputFolderPath))
{
Directory.CreateDirectory(outputFolderPath);
}
var locationExtraction = new NetworkLocationExtraction(parser);
var stringsCount = new Dictionary <string, int>(StringComparer.OrdinalIgnoreCase);
var normalizedLocationType = typeof(NormalizedLocation);
var properties = normalizedLocationType.GetProperties(BindingFlags.Public | BindingFlags.Instance);
PropertyInfo targetProperty = null;
foreach (var property in properties)
{
if (property.Name == propertyName)
{
targetProperty = property;
}
}
if (targetProperty == null)
{
throw new ArgumentNullException("targetProperty");
}
foreach (var network in locationExtraction.ExtractNetworksWithLocations(inputFilePath, inputFilePath))
{
if (network.Id != null)
{
var rawPropertyValue = targetProperty.GetValue(network.Location);
if (rawPropertyValue != null)
{
var value = (string)rawPropertyValue;
int currentCount;
if (!stringsCount.TryGetValue(value, out currentCount))
{
currentCount = 0;
}
currentCount++;
stringsCount[value] = currentCount;
}
}
//// TODO: Else log
}
using (var outputFile = new StreamWriter(outputFilePath))
{
foreach (var entry in stringsCount)
{
// No need to sanitize entry.Value since it's a number
outputFile.WriteLine(string.Format(CultureInfo.InvariantCulture, "{0}\t{1}", TsvUtils.ReplaceAndTrimIllegalCharacters(entry.Key, removeDoubleQuotes: true), entry.Value));
}
}
}
public void ColumnsPerTypeToTsv(string inputFilePath, string outputFilePath)
{
var columnsPerTypes = this.Parser.ColumnsPerType(inputFilePath);
using (var outputFile = new StreamWriter(outputFilePath))
{
foreach (var columnsPerType in columnsPerTypes)
{
foreach (var column in columnsPerType.Value)
{
outputFile.WriteLine(string.Format(CultureInfo.InvariantCulture, "{0}\t{1}", TsvUtils.ReplaceAndTrimIllegalCharacters(columnsPerType.Key, removeDoubleQuotes: true), TsvUtils.ReplaceAndTrimIllegalCharacters(column, removeDoubleQuotes: true)));
}
}
}
}
public void TypeToFieldDistinctOcc(string inputFolderPath, string outputFilePath)
{
var globalTypeToFieldDistinctOcc = new Dictionary <string, Dictionary <string, int> >();
foreach (var file in Directory.GetFiles(inputFolderPath))
{
this.Parser.ResetFieldStats();
var localTypeToFieldDistinctOcc = this.Parser.TypeToFieldDistinctOcc(file);
this.MergeIntoGlobalTypeCounts(globalTypeToFieldDistinctOcc, localTypeToFieldDistinctOcc);
}
using (var outputFile = new StreamWriter(outputFilePath))
{
foreach (var entry in globalTypeToFieldDistinctOcc)
{
var type = entry.Key;
var fieldOcc = entry.Value;
foreach (var occEntry in fieldOcc)
{
outputFile.WriteLine(string.Format(CultureInfo.InvariantCulture, "{0}\t{1}\t{2}", TsvUtils.ReplaceAndTrimIllegalCharacters(type, removeDoubleQuotes: true), occEntry.Key, occEntry.Value.ToString()));
}
}
}
}
public void ColumnsPerTypeToTsv(string inputFolderPath, string outputFilePath)
{
var globalColumnsPerType = new Dictionary <string, List <string> >();
foreach (var file in Directory.GetFiles(inputFolderPath))
{
this.Parser.ResetFieldStats();
var localColumnsPerTypes = this.Parser.ColumnsPerType(file);
this.MergeIntoGlobalColumnsPerType(globalColumnsPerType, localColumnsPerTypes);
}
using (var outputFile = new StreamWriter(outputFilePath))
{
foreach (var entry in globalColumnsPerType)
{
var recordType = entry.Key;
var recordColumns = entry.Value;
foreach (var recordColumn in recordColumns)
{
outputFile.WriteLine(string.Format(CultureInfo.InvariantCulture, "{0}\t{1}", TsvUtils.ReplaceAndTrimIllegalCharacters(recordType, removeDoubleQuotes: true), TsvUtils.ReplaceAndTrimIllegalCharacters(recordColumn, removeDoubleQuotes: true)));
}
}
}
}
public void TypeToFieldDistinctOcc(string inputFilePath, string outputFilePath)
{
var typeToFieldDistinctOcc = this.Parser.TypeToFieldDistinctOcc(inputFilePath);
using (var outputFile = new StreamWriter(outputFilePath))
{
foreach (var entry in typeToFieldDistinctOcc)
{
var key = entry.Key;
var fieldOcc = entry.Value;
foreach (var fieldOccEntry in fieldOcc)
{
outputFile.WriteLine(string.Format(CultureInfo.InvariantCulture, "{0}\t{1}\t{2}", TsvUtils.ReplaceAndTrimIllegalCharacters(key, removeDoubleQuotes: true), TsvUtils.ReplaceAndTrimIllegalCharacters(fieldOccEntry.Value.ToString(), removeDoubleQuotes: true), fieldOccEntry.Value.ToString()));
}
}
}
}
public void TypeCountsToTsv(string inputFilePath, string outputFilePath)
{
var typeCounts = this.Parser.TypeCounts(inputFilePath);
using (var outputFile = new StreamWriter(outputFilePath))
{
foreach (var entry in typeCounts)
{
outputFile.WriteLine(string.Format(CultureInfo.InvariantCulture, "{0}\t{1}", TsvUtils.ReplaceAndTrimIllegalCharacters(entry.Key, removeDoubleQuotes: true), TsvUtils.ReplaceAndTrimIllegalCharacters(entry.Value.ToString(), removeDoubleQuotes: true)));
}
}
}
