public static bool IsPlainText(string fileName)
{
var fileInfo = new FileInfo(fileName);
if (fileInfo.Length < 20)
{
return(false); // too short to be plain text
}
if (fileInfo.Length > 5000000)
{
return(false); // too large to be plain text
}
var enc = LanguageAutoDetect.GetEncodingFromFile(fileName);
var s = ReadAllTextShared(fileName, enc);
int numberCount = 0;
int letterCount = 0;
int len = s.Length;
for (int i = 0; i < len; i++)
{
var ch = s[i];
if (char.IsLetter(ch) || " -,.!?[]()\r\n".Contains(ch))
{
letterCount++;
}
else if (char.IsControl(ch) && ch != '\t') // binary found
{
return(false);
}
else if (CharUtils.IsDigit(ch))
{
numberCount++;
}
}
if (len < 100)
{
return(numberCount < 5 && letterCount > 20);
}
var numberPatternMatches = new Regex(@"\d+[.:,; -]\d+").Matches(s);
if (numberPatternMatches.Count > 30)
{
return(false); // looks like time codes
}
var largeBlocksOfLargeNumbers = new Regex(@"\d{3,8}").Matches(s);
if (largeBlocksOfLargeNumbers.Count > 30)
{
return(false); // looks like time codes
}
if (len < 1000 && largeBlocksOfLargeNumbers.Count > 10)
{
return(false); // looks like time codes
}
var partsWithMoreThan100CharsOfNonNumbers = new Regex(@"[^\d]{150,100000}").Matches(s);
if (partsWithMoreThan100CharsOfNonNumbers.Count > 10)
{
return(true); // looks like text
}
var numberThreshold = len * 0.015 + 25;
var letterThreshold = len * 0.8;
return(numberCount < numberThreshold && letterCount > letterThreshold);
}
public SubtitleFormat LoadSubtitle(string fileName, out Encoding encoding, Encoding useThisEncoding, bool batchMode)
{
FileName = fileName;
_paragraphs = new List <Paragraph>();
var lines = new List <string>();
StreamReader sr;
if (useThisEncoding != null)
{
try
{
sr = new StreamReader(fileName, useThisEncoding);
}
catch (Exception exception)
{
System.Diagnostics.Debug.WriteLine(exception.Message);
encoding = Encoding.UTF8;
return(null);
}
}
else
{
try
{
sr = new StreamReader(fileName, LanguageAutoDetect.GetEncodingFromFile(fileName), true);
}
catch
{
try
{
Stream fs = new FileStream(fileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite);
sr = new StreamReader(fs);
}
catch (Exception exception)
{
System.Diagnostics.Debug.WriteLine(exception.Message);
encoding = Encoding.UTF8;
return(null);
}
}
}
encoding = sr.CurrentEncoding;
while (!sr.EndOfStream)
{
lines.Add(sr.ReadLine());
}
sr.Close();
foreach (SubtitleFormat subtitleFormat in SubtitleFormat.AllSubtitleFormats)
{
if (subtitleFormat.IsMine(lines, fileName))
{
Header = null;
subtitleFormat.BatchMode = batchMode;
subtitleFormat.LoadSubtitle(this, lines, fileName);
_format = subtitleFormat;
_wasLoadedWithFrameNumbers = _format.IsFrameBased;
if (_wasLoadedWithFrameNumbers)
{
CalculateTimeCodesFromFrameNumbers(Configuration.Settings.General.CurrentFrameRate);
}
return(subtitleFormat);
}
}
if (useThisEncoding == null)
{
return(LoadSubtitle(fileName, out encoding, Encoding.Unicode));
}
return(null);
}
public SubtitleFormat LoadSubtitle(string fileName, out Encoding encoding, Encoding useThisEncoding, bool batchMode, double?sourceFrameRate = null, bool loadSubtitle = true)
{
FileName = fileName;
Paragraphs = new List <Paragraph>();
StreamReader sr;
if (useThisEncoding != null)
{
try
{
sr = new StreamReader(fileName, useThisEncoding);
}
catch (Exception exception)
{
System.Diagnostics.Debug.WriteLine(exception.Message);
encoding = Encoding.UTF8;
return(null);
}
}
else
{
try
{
sr = new StreamReader(fileName, LanguageAutoDetect.GetEncodingFromFile(fileName), true);
}
catch
{
try
{
var fs = new FileStream(fileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite);
sr = new StreamReader(fs);
}
catch (Exception exception)
{
System.Diagnostics.Debug.WriteLine(exception.Message);
encoding = Encoding.UTF8;
return(null);
}
}
}
encoding = sr.CurrentEncoding;
var lines = sr.ReadToEnd().SplitToLines();
sr.Close();
var ext = Path.GetExtension(fileName).ToLowerInvariant();
foreach (var subtitleFormat in SubtitleFormat.AllSubtitleFormats.Where(p => p.Extension == ext && !p.Name.StartsWith("Unknown", StringComparison.Ordinal)))
{
if (subtitleFormat.IsMine(lines, fileName))
{
return(FinalizeFormat(fileName, batchMode, sourceFrameRate, lines, subtitleFormat, loadSubtitle));
}
}
foreach (var subtitleFormat in SubtitleFormat.AllSubtitleFormats.Where(p => p.Extension != ext || p.Name.StartsWith("Unknown", StringComparison.Ordinal)))
{
if (subtitleFormat.IsMine(lines, fileName))
{
return(FinalizeFormat(fileName, batchMode, sourceFrameRate, lines, subtitleFormat, loadSubtitle));
}
}
if (useThisEncoding == null)
{
return(LoadSubtitle(fileName, out encoding, Encoding.Unicode));
}
return(null);
}