static void Main(string[] args)
{
var parser = new CsvParser();
var config = BuildConfiguration();
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
var encoding = EncodingDetector.GetEncoding(config["data-file-path"]);
var data = parser.ParseFile(config["data-file-path"], encoding);
if (parser.ErrorMessage != null)
{
Console.WriteLine("Parsing failed");
Console.WriteLine(parser.ErrorMessage);
}
var report = DataAnalyser.Analyze(data, Convert.ToUInt32(config["min-number-for-adv-stat"]));
using (var stream = new FileStream("report.txt", FileMode.Create))
{
TextReporter.ToStream(stream, report);
}
var reporter = new ExcelReport.ExcelReporter();
using (var stream = new FileStream("report.xlsx", FileMode.Create))
{
reporter.ToStream(stream, report);
}
}
public MovieCollection ImportMoviesFromDumpFiles(string[] filesToRead, MovieCollection movieCollection)
{
var movieList = new List <Movie>();
foreach (var txtFile in filesToRead)
{
if (File.Exists(txtFile) && txtFile.Contains(".txt"))
{
var fileName = Path.GetFileNameWithoutExtension(txtFile);
var Encoding = EncodingDetector.DetectTextFileEncoding(txtFile);
var myFile = new StreamReader(txtFile, Encoding);
var moviesString = myFile.ReadToEnd();
myFile.Close();
moviesString = moviesString.Replace("last-modified |ext |size|name|location", string.Empty);
var moviesLines = moviesString.Split(new string[] { Environment.NewLine }, StringSplitOptions.RemoveEmptyEntries);
OnProgressUpdate?.Invoke(ImportProgressType.NewFile, fileName, moviesLines.Count());
foreach (var line in moviesLines)
{
if (line != "done" && !string.IsNullOrWhiteSpace(line))
{
var movie = FileHelper.GetMovieFromLine(line, fileName);
if (movie != null)
{
movieCollection.AddMovie(movie);
}
}
OnProgressUpdate?.Invoke(ImportProgressType.NewLine, string.Empty, 0);
}
}
}
return(movieCollection);
}
private bool CheckCUE()
{
State = FileState.ValidFile;
Encode = EncodingDetector.GetEncoding(FullPath, out _confidence);
if (Encode != "UTF-8")
{
State = FileState.InValidEncode;
return(false);
}
using (var fs = File.OpenRead(FullPath))
{
var buffer = new byte[3];
fs.Read(buffer, 0, 3);
if (buffer[0] == 0xEF && buffer[1] == 0xBB && buffer[2] == 0xBF)
{
if (!CueCurer.CueMatchCheck(this))
{
State = FileState.InValidCue;
}
return(true);
}
}
State = FileState.NonUTF8WBOM;
return(false);
}
private void RunTest(EncodingDetector.Options encDetectorOptions)
{
int totalFiles = 0;
int numFoundEncodings = 0;
foreach (string filePath in Directory.EnumerateFiles(_tempDir, "*.*", SearchOption.AllDirectories))
{
var sampleBytes = Utils.ReadFileContentSample(filePath);
Encoding encoding = EncodingDetector.Detect(sampleBytes, encDetectorOptions);
totalFiles++;
if (encoding != null)
{
numFoundEncodings++;
}
WriteToConsole(Path.GetFileName(filePath) + ": " + encoding);
if (totalFiles > 10)
{
break;
}
}
Console.WriteLine("Found Encoding in:" + numFoundEncodings + " out of " + totalFiles);
}
/// <summary>
/// Загрузить описание базовых объектов из файла
/// </summary>
/// <param name="pathToFile">Путь к файлу с базовыми объектами</param>
/// <returns>Описание</returns>
private string LoadBaseTechObjectsDescription(string pathToFile)
{
var reader = new StreamReader(pathToFile,
EncodingDetector.DetectFileEncoding(pathToFile));
string readedDescription = reader.ReadToEnd();
return(readedDescription);
}
public void IsXmlEncoded_ValidInput_ReturnsTrue()
{
// Arrange
var input = "MSH.1>MSH.2>";
// Act / Assert
Assert.IsTrue(EncodingDetector.IsXmlEncoded(input));
}
void IncludeFile(string fileName, List <TokenExpression> tokens)
{
var encoding = EncodingDetector.DetectEncoding(fileName);
var trBase = new StreamReader(fileName, encoding);
var tr = new MyTextReader2(trBase, fileName);
IncludeFile(tr, tokens, fileName);
}
public void AssertXmlEncoded_ValidInput_NoExceptionThrown()
{
// Arrange
var input = "MSH.1>MSH.2>";
// Arrange / Act / Assert
Assert.DoesNotThrow(
() => EncodingDetector.AssertXmlEncoded(input));
}
/// <summary>
/// Загрузить LUA-скрипт из файла
/// </summary>
/// <param name="path">Путь к файлу скрипта</param>
private void LoadScript(string path)
{
var reader = new StreamReader(path,
EncodingDetector.DetectFileEncoding(path));
string script = reader.ReadToEnd();
reader.Close();
lua.DoString(script);
}
public void AssertEr7Encoded_4thCharacterOfgivenSegmentNotDelimiter_ThrowsInvalidOperationException()
{
// Arrange
var input = "MSH|^~\\&||\rUNKNOWN|";
// Act / Assert
Assert.Throws <InvalidOperationException>(
() => EncodingDetector.AssertEr7Encoded(input));
}
public void IsUTF8Test()
{
foreach (var item in Directory.GetFiles(@"..\..\[Encode Sample]"))
{
float confindence;
string encode = EncodingDetector.GetEncoding(item, out confindence);
Console.WriteLine($"{Path.GetFileName(item)}: {encode == "UTF-8"} confidence: {confindence:F3}");
}
}
public void IsEr7Encoded_InvalidInput_ReturnsTrue()
{
// Arrange
var input = "MSH|^~\\&|LABGL1||DMCRES||19951002185200||ADT^A01|LABGL1199510021852632|P|2.2\r"
+ "PID|||T12345||TEST^PATIENT^P||19601002|M||||||||||123456\r"
+ "PV1|||NER|||||||GSU||||||||E||||||||||||||||||||||||||19951002174900|19951006\r";
// Act / Assert
Assert.IsTrue(EncodingDetector.IsEr7Encoded(input));
}
public void TestDetectingCodepages(int codepage)
{
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
var stream = Common.CreateStreamFromText(
Common.TestResultsDataString,
Encoding.GetEncoding(codepage));
var encoding = EncodingDetector.GetEncoding(stream);
Assert.Equal(codepage, encoding.CodePage);
}
public void AssertEr7Encoded_ValidInput_NoExceptionThrown()
{
// Arrange
var input = "MSH|^~\\&|LABGL1||DMCRES||19951002185200||ADT^A01|LABGL1199510021852632|P|2.2\r"
+ "PID|||T12345||TEST^PATIENT^P||19601002|M||||||||||123456\r"
+ "PV1|||NER|||||||GSU||||||||E||||||||||||||||||||||||||19951002174900|19951006\r";
// Act / Assert
Assert.DoesNotThrow(
() => EncodingDetector.AssertEr7Encoded(input));
}
public void TestDetectNoPreamble()
{
using (var stream = new MemoryStream())
using (var writer = new StreamWriter(stream, Encoding.ASCII))
{
writer.Write("Controversial opinion: White House Down is a better movie than Olympus has Fallen, ikr");
writer.Flush();
stream.Position = 0;
var detector = new EncodingDetector();
detector.TryFindEncoding(stream).Should().Be(null, "because there's no preamble to detect ASCII and the detector shall not make any guesses and give up");
}
}
public void EncodeTest()
{
float confindence;
foreach (var item in Directory.GetFiles(@"..\..\[Encode Sample]"))
{
Console.WriteLine($"{Path.GetFileName(item)}: {EncodingDetector.GetEncoding(item, out confindence)} ({confindence:F3})");
}
foreach (var item in Directory.GetFiles(@"..\..\[Encoding All Star]"))
{
Console.WriteLine($"{Path.GetFileName(item)}: {EncodingDetector.GetEncoding(item, out confindence)} ({confindence:F3})");
}
}
/// <summary>
/// Прочитать shared.lua в список для манипуляций
/// </summary>
/// <param name="projName">Имя проекта для поиска модели</param>
/// <param name="pathToSharedFile">Путь к файлу shared.lua</param>
private void ReadModelSharedFileToList(string projName,
string pathToSharedFile)
{
IProjectModel model = interprojectExchange.Models
.Where(x => x.ProjectName == projName)
.FirstOrDefault();
if (model != null)
{
model.SharedFileAsStringList = File
.ReadAllLines(pathToSharedFile,
EncodingDetector.DetectFileEncoding(pathToSharedFile))
.ToList();
}
}
/// <summary>
/// Detects encoding of a filestream when BOM is present
/// </summary>
/// <param name="fileName"></param>
public static void ExtractEncodingByBOM(string fileName)
{
//ExStart:ExtractEncodingByBOM
try
{
EncodingDetector detector = new EncodingDetector(Encoding.GetEncoding(1251));
//get file actual path
String filePath = Common.GetFilePath(fileName);
Stream stream = new FileStream(filePath, FileMode.Open);
Console.WriteLine(detector.Detect(stream));
}
catch (Exception ex)
{
Console.WriteLine(ex.Message);
}
//ExEnd:ExtractEncodingByBOM
}
public void TestDetect([ValueSource(nameof(EncodingsWithPreamble))] Encoding encoding)
{
var preamble = encoding.GetPreamble();
preamble.Length.Should().BeGreaterOrEqualTo(1, "because this test doesn't make much sense if the encoding has no preamble");
using (var stream = new MemoryStream())
using (var writer = new StreamWriter(stream, encoding))
{
writer.Write("Controversial opinion: White House Down is a better movie than Olympus has Fallen, ikr");
writer.Flush();
stream.Position = 0;
var detector = new EncodingDetector();
detector.TryFindEncoding(stream).Should().Be(encoding);
}
}
public bool Load()
{
if (!File.Exists(FileName))
{
return(false);
}
try
{
Sections.Clear();
using (var fileData = new StreamReader(FileName, FileEncoding, true))
{
string firstLine = null;
while (string.IsNullOrEmpty(firstLine) && !fileData.EndOfStream)
{
firstLine = fileData.ReadLine();
}
if (!FileEncoding.Equals(Encoding.Unicode) && !FileEncoding.Equals(Encoding.BigEndianUnicode) &&
EncodingDetector.IsUnicode(firstLine))
{
FileEncoding = Encoding.Unicode;
}
else
{
FileEncoding = fileData.CurrentEncoding;
}
fileData.Close();
}
using (var fileData = new StreamReader(FileName, FileEncoding, true))
{
LoadContent(fileData);
fileData.Close();
}
return(true);
}
catch (IOException ex)
{
System.Diagnostics.Trace.TraceError(ex.Message);
return(false);
}
}
public ActionResult ExtractDocumentEndocing([FromBody] string fileName)
{
List <string> extractedText = new List <string>();
ExtractorFactory factory = new ExtractorFactory();
string filePath = Server.MapPath("../App_Data//Uploads//" + fileName);
try
{
EncodingDetector detector = new EncodingDetector(Encoding.GetEncoding(1251));
Stream stream = new FileStream(filePath, FileMode.Open, FileAccess.Read, FileShare.ReadWrite);
extractedText.Add(detector.Detect(stream).ToString());
}
catch (Exception ex)
{
extractedText.Add("File Format not supported");
}
return(Json(extractedText, JsonRequestBehavior.AllowGet));
}
/// <summary>
/// Чтение Shared файла альтернативного проекта
/// </summary>
/// <param name="pathToProjectsDir">Путь к каталогу с проектами</param>
/// <param name="projName">Имя проекта</param>
/// <returns></returns>
public void LoadAdvancedProjectSharedLuaData(string pathToProjectsDir,
string projName)
{
string pathToSharedFile = Path.Combine(pathToProjectsDir, projName,
signalsFile);
if (File.Exists(pathToSharedFile))
{
var reader = new StreamReader(pathToSharedFile,
EncodingDetector.DetectFileEncoding(pathToSharedFile));
string sharedInfo = reader.ReadToEnd();
reader.Close();
lua.DoString(sharedInfo);
// Функция из Lua
lua.DoString("init_advanced_project_shared_lua()");
ReadModelSharedFileToList(projName, pathToSharedFile);
}
}
private void DetectEncodingSyncAction(DetectEncodingFileDto dto)
{
string stopWatchKey = "DetectEncodingSyncAction_" + Thread.CurrentThread.ManagedThreadId;
StopWatch.Start(stopWatchKey);
//First try BOM detection and Unicode detection using Klerks Soft encoder
//stream.Seek(0, SeekOrigin.Begin);
Encoding encoding = EncodingDetector.Detect(dto.SampleBytes, EncodingDetector.Options.MLang);
//Encoding encoding = null;
//Encoding[] detected;
//try
//{
// detected = EncodingTools.DetectInputCodepages(dto.SampleBytes, 1);
// if (detected.Length > 0)
// encoding = detected[0];
//}
//catch (COMException ex)
//{
// // return default codepage on error
//}
lock (lockObj)
{
_totalFiles++;
if (encoding != null)
{
_numFoundEncodings++;
}
}
//WriteToConsole(dto.FilePath + ": " + encoding);
StopWatch.Stop(stopWatchKey);
}
public static Event?Recode(FileContext context)
{
byte[] content = File.ReadAllBytes(context.FullFileName);
Encoding encoding = EncodingDetector.Detect(content);
if (encoding != FileRecoder.win1252)
{
return(null);
}
int length = (int)(content.Length * 1.5);
if (FileRecoder.resultContent.Length < length)
{
FileRecoder.resultContent = new byte[length];
}
using (MemoryStream mem = new(FileRecoder.resultContent))
{
foreach (byte b in content)
{
if (b >= 0x80)
{
byte[] buffer = FileRecoder.charmap1[b - 0x80];
mem.Write(buffer, 0, buffer.Length);
}
else
{
mem.WriteByte(b);
}
}
length = (int)mem.Position;
}
using FileStream stream = File.OpenWrite(context.FullFileName);
stream.Write(FileRecoder.resultContent, 0, length);
return(null);
}
/// <summary>
/// Загрузка devices.lua файла проекта
/// </summary>
/// <param name="pathToProjectsDir"></param>
/// <param name="projName"></param>
private void LoadDevicesFile(string pathToProjectsDir,
string projName)
{
string pathToDevices = Path.Combine(pathToProjectsDir, projName,
deviceDescriptionFile);
if (File.Exists(pathToDevices))
{
var reader = new StreamReader(pathToDevices,
EncodingDetector.DetectFileEncoding(pathToDevices));
string devicesInfo = reader.ReadToEnd();
reader.Close();
lua.DoString(devicesInfo);
// Функция из Lua
lua.DoString("system.init_dev_names()");
}
else
{
form.ShowErrorMessage($"Не найден файл main.devices.lua " +
$"проекта \"{projName}\"");
}
}
/// <summary>
/// Чтение информации о ПЛК из main.io.lua
/// </summary>
/// <param name="pathToProjectsDir">Путь к папке с проектами</param>
/// <param name="projName">Имя проекта</param>
/// <returns></returns>
private void LoadMainIOData(string pathToProjectsDir,
string projName)
{
string pathToIOFile = Path.Combine(pathToProjectsDir, projName,
devicesAndPLCFile);
if (File.Exists(pathToIOFile))
{
var reader = new StreamReader(pathToIOFile,
EncodingDetector.DetectFileEncoding(pathToIOFile));
string ioInfo = reader.ReadToEnd();
reader.Close();
lua.DoString(ioInfo);
// Функция из Lua
lua.DoString("init_io_file()");
}
else
{
MessageBox.Show($"Не найден файл main.io.lua проекта" +
$" \"{projName}\"", "Ошибка", MessageBoxButtons.OK,
MessageBoxIcon.Error);
}
}
/// <summary>
/// Загрузка описание базовых объектов
/// </summary>
/// <returns>Описание</returns>
private string LoadBaseTechObjectsDescription()
{
var fileName = "sys_base_objects_description.lua";
var pathToFile = Path.Combine(
ProjectManager.GetInstance().SystemFilesPath, fileName);
if (!File.Exists(pathToFile))
{
string template = EasyEPlanner.Properties.Resources
.ResourceManager
.GetString("SysBaseObjectsDescriptionPattern");
File.WriteAllText(pathToFile, template,
EncodingDetector.UTF8);
MessageBox.Show("Файл с описанием базовых объектов не найден." +
" Будет создан пустой файл (без описания).", "Ошибка",
MessageBoxButtons.OK, MessageBoxIcon.Error);
}
var reader = new StreamReader(pathToFile,
EncodingDetector.DetectFileEncoding(pathToFile));
string readedDescription = reader.ReadToEnd();
return(readedDescription);
}
private async Task <Response> ParseFileText(string fileName, string folderName)
{
string logMsg = "ControllerName: GroupDocsParserController FileName: " + fileName + " FolderName: " + folderName;
try
{
return(await ProcessTask(fileName, folderName, ".txt", false, "", delegate(string inFilePath, string outPath, string zipOutFolder)
{
EncodingDetector detector = new EncodingDetector(Encoding.GetEncoding(1251));
if (!Directory.Exists(zipOutFolder))
{
Directory.CreateDirectory(zipOutFolder);
}
using (Stream stream = new FileStream(inFilePath, FileMode.Open)) {
System.IO.File.WriteAllText(outPath, "Encoding: " + detector.Detect(stream, true) + Environment.NewLine);
}
ExtractorFactory factory = new ExtractorFactory();
MetadataExtractor metadataExtractor = factory.CreateMetadataExtractor(inFilePath);
if (metadataExtractor != null)
{
MetadataCollection metadataCollection = metadataExtractor.ExtractMetadata(inFilePath);
System.IO.File.AppendAllText(outPath, Environment.NewLine + "Metadata:" + Environment.NewLine);
foreach (string key in metadataCollection.Keys)
{
System.IO.File.AppendAllText(outPath, string.Format("{0} = {1}", key, metadataCollection[key]) + Environment.NewLine);
}
}
System.IO.File.AppendAllText(outPath, Environment.NewLine + "Parsed content:" + Environment.NewLine);
string fileExt = Path.GetExtension(fileName).Substring(1).ToLower();
if (GetFormatType(fileExt) == FormatType.Excel)
{
CellsTextExtractor extractor = new CellsTextExtractor(inFilePath);
extractor.ExtractMode = ExtractMode.Standard;
for (int sheetIndex = 0; sheetIndex < extractor.SheetCount; sheetIndex++)
{
System.IO.File.AppendAllText(outPath, Environment.NewLine + "Sheet # " + extractor.SheetCount + Environment.NewLine);
System.IO.File.AppendAllText(outPath, extractor.ExtractSheet(sheetIndex));
}
}
else
{
TextExtractor textExtractor = factory.CreateFormattedTextExtractor(inFilePath);
if (textExtractor == null)
{
textExtractor = factory.CreateTextExtractor(inFilePath);
}
System.IO.File.AppendAllText(outPath, textExtractor.ExtractAll());
}
}));
}
catch (Exception exc)
{
return(new Response {
FileName = fileName, FolderName = folderName, OutputType = "txt", Status = exc.Message, StatusCode = 500, Text = exc.ToString()
});
}
}
public void IsXmlEncoded_InvalidInput_ReturnsFalse(string input)
{
// Arrange / Act / Assert
Assert.IsFalse(EncodingDetector.IsXmlEncoded(input));
}
public void AssertXmlEncoded_InValidInput_ThrowsArgumentException(string input)
{
// Arrange / Act / Assert
Assert.Throws <ArgumentException>(
() => EncodingDetector.AssertXmlEncoded(input));
}