Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-23 11:42:36 +01:00
Code Issues Projects Releases Wiki Activity
f0aab41c56
SubtitleEdit/libse/UknownFormatImporter.cs

779 lines
35 KiB
C#
Raw Normal View History

Some line endings to CRLF
2016-02-08 21:11:03 +01:00
using Nikse.SubtitleEdit.Core.SubtitleFormats;
using System;
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
using System.Collections.Generic;
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
using System.Text;
using System.Text.RegularExpressions;
namespace Nikse.SubtitleEdit.Core
{
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
/// <summary>
/// Generic subtitle format parser
/// </summary>
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
public class UknownFormatImporter
{
private static readonly char[] ExpectedSplitChars = { '.', ',', ';', ':' };
public bool UseFrames { get; set; }
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
public Subtitle AutoGuessImport(List<string> lines)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
var subtitle = ImportTimeCodesOnSameSeperateLine(lines);
if (subtitle.Paragraphs.Count < 2)
subtitle = ImportTimeCodesAndTextOnSameLineOnlySpaceAsSeparator(lines);
var subTcAndTextOnSameLine = ImportTimeCodesAndTextOnSameLine(lines);
if (subTcAndTextOnSameLine.Paragraphs.Count > subtitle.Paragraphs.Count)
subtitle = subTcAndTextOnSameLine;
var subTcOnAloneLines = ImportTimeCodesOnAloneLines(lines);
if (subTcOnAloneLines.Paragraphs.Count > subtitle.Paragraphs.Count)
subtitle = subTcOnAloneLines;
if (subtitle.Paragraphs.Count < 2)
{
subtitle = ImportTimeCodesInFramesOnSameSeperateLine(lines);
if (subtitle.Paragraphs.Count < 2)
{
subtitle = ImportTimeCodesInFramesAndTextOnSameLine(lines);
}
}
if (subtitle.Paragraphs.Count > 1)
Added generic json subtitle parser
2017-02-11 17:02:22 +01:00
{
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
CleanUp(subtitle);
Added generic json subtitle parser
2017-02-11 17:02:22 +01:00
}
if (subtitle.Paragraphs.Count < 2)
{
var jsonSubtitle = new UknownFormatImporterJson().AutoGuessImport(lines);
if (jsonSubtitle != null && jsonSubtitle.Paragraphs.Count > 2)
subtitle = jsonSubtitle;
}
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
if (subtitle.Paragraphs.Count == 0 && lines.Count == 1 && lines[0].Contains(" --> "))
Allow import of SRT without line breaks
2017-11-19 13:32:18 +01:00
{
subtitle = ImportSubtitleWithNoLineBreaks(lines[0]);
}
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
return subtitle;
}
private static void CleanUp(Subtitle subtitle)
{
foreach (Paragraph p in subtitle.Paragraphs)
{
p.Text = p.Text.Replace("<html>", string.Empty);
p.Text = p.Text.Replace("</html>", string.Empty);
p.Text = p.Text.Replace("<div>", string.Empty);
p.Text = p.Text.Replace("</div>", string.Empty);
p.Text = p.Text.Replace("<body>", string.Empty);
p.Text = p.Text.Replace("</body>", string.Empty);
p.Text = p.Text.Replace("<tt>", string.Empty);
p.Text = p.Text.Replace("</tt>", string.Empty);
p.Text = p.Text.Replace("<tr>", string.Empty);
p.Text = p.Text.Replace("</tr>", string.Empty);
p.Text = p.Text.Replace("<td>", string.Empty);
p.Text = p.Text.Replace("</td>", string.Empty);
p.Text = p.Text.Replace("<table>", string.Empty);
p.Text = p.Text.Replace("</table>", string.Empty);
p.Text = p.Text.Replace("<br>", Environment.NewLine);
p.Text = p.Text.Replace("<br/>", Environment.NewLine);
p.Text = p.Text.Replace("<br />", Environment.NewLine);
p.Text = p.Text.Replace("&lt;", "<");
p.Text = p.Text.Replace("&gt;", ">");
p.Text = p.Text.Replace(" ", " ");
p.Text = p.Text.Replace(" ", " ");
p.Text = p.Text.Replace(" ", " ");
p.Text = p.Text.Replace("|", Environment.NewLine).Replace("<p>", Environment.NewLine).Replace("</p>", Environment.NewLine).Trim();
p.Text = p.Text.Replace(Environment.NewLine + Environment.NewLine, Environment.NewLine).Trim();
p.Text = p.Text.Replace(Environment.NewLine + Environment.NewLine, Environment.NewLine).Trim();
}
subtitle.RemoveEmptyLines();
}
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
private Subtitle ImportTimeCodesInFramesAndTextOnSameLine(List<string> lines)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
var regexTimeCodes1 = new Regex(@"\d+", RegexOptions.Compiled);
Paragraph p = null;
var subtitle = new Subtitle();
var sb = new StringBuilder();
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
for (int idx = 0; idx < lines.Count; idx++)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
string line = lines[idx];
var matches = regexTimeCodes1.Matches(line);
if (matches.Count >= 2)
{
string start = matches[0].ToString();
string end = matches[1].ToString();
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
try
{
if (UseFrames)
{
p.StartFrame = int.Parse(start);
p.EndFrame = int.Parse(end);
p.CalculateTimeCodesFromFrameNumbers(Configuration.Settings.General.CurrentFrameRate);
}
else
{
p.StartTime.TotalMilliseconds = double.Parse(start);
p.EndTime.TotalMilliseconds = double.Parse(end);
}
}
catch
{
p = null;
}
if (matches[0].Index < 9)
line = line.Remove(0, matches[0].Index);
line = line.Replace(matches[0].ToString(), string.Empty);
line = line.Replace(matches[1].ToString(), string.Empty);
line = line.Trim();
if (line.StartsWith("}{}", StringComparison.Ordinal) || line.StartsWith("][]", StringComparison.Ordinal))
line = line.Remove(0, 3);
line = line.Trim();
}
if (p != null && line.Length > 1)
{
sb.AppendLine(line.Trim());
if (sb.Length > 200)
return new Subtitle();
}
}
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
subtitle.Renumber();
return subtitle;
}
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
private Subtitle ImportTimeCodesInFramesOnSameSeperateLine(List<string> lines)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
Paragraph p = null;
var subtitle = new Subtitle();
var sb = new StringBuilder();
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
for (int idx = 0; idx < lines.Count; idx++)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
string line = lines[idx];
string lineWithPerhapsOnlyNumbers = GetLineWithPerhapsOnlyNumbers(line);
bool allNumbers = lineWithPerhapsOnlyNumbers.Length > 0;
foreach (char c in lineWithPerhapsOnlyNumbers)
{
if (!char.IsDigit(c))
allNumbers = false;
}
if (allNumbers && lineWithPerhapsOnlyNumbers.Length > 2)
{
string[] arr = line.Replace('-', ' ').Replace('>', ' ').Replace('{', ' ').Replace('}', ' ').Replace('[', ' ').Replace(']', ' ').Trim().Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
if (arr.Length == 2)
{
string[] start = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
string[] end = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (start.Length == 1 && end.Length == 1)
{
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
try
{
if (UseFrames)
{
p.StartFrame = int.Parse(start[0]);
p.EndFrame = int.Parse(end[0]);
p.CalculateTimeCodesFromFrameNumbers(Configuration.Settings.General.CurrentFrameRate);
}
else
{
p.StartTime.TotalMilliseconds = double.Parse(start[0]);
p.EndTime.TotalMilliseconds = double.Parse(end[0]);
}
}
catch
{
p = null;
}
}
}
else if (arr.Length == 3)
{
string[] start = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
string[] end = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
string[] duration = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (end.Length == 1 && duration.Length == 1)
{
start = end;
end = duration;
}
if (start.Length == 1 && end.Length == 1)
{
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
try
{
if (UseFrames)
{
p.StartFrame = int.Parse(start[0]);
p.EndFrame = int.Parse(end[0]);
p.CalculateTimeCodesFromFrameNumbers(Configuration.Settings.General.CurrentFrameRate);
}
else
{
p.StartTime.TotalMilliseconds = double.Parse(start[0]);
p.EndTime.TotalMilliseconds = double.Parse(end[0]);
}
}
catch
{
p = null;
}
}
}
}
if (p != null && !allNumbers && line.Length > 1)
{
line = line.Trim();
if (line.StartsWith("}{}", StringComparison.Ordinal) || line.StartsWith("][]", StringComparison.Ordinal))
line = line.Remove(0, 3);
sb.AppendLine(line.Trim());
}
}
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
subtitle.CalculateTimeCodesFromFrameNumbers(Configuration.Settings.General.CurrentFrameRate);
subtitle.Renumber();
return subtitle;
}
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
private static Subtitle ImportTimeCodesOnAloneLines(List<string> lines)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
Paragraph p = null;
var subtitle = new Subtitle();
var sb = new StringBuilder();
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
char[] splitChars = { ' ', '\t', '-', '>', '<', '{', '}', '[', ']' };
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
for (int idx = 0; idx < lines.Count; idx++)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
string line = lines[idx];
string lineWithPerhapsOnlyNumbers = GetLineWithPerhapsOnlyNumbers(line);
bool allNumbers = lineWithPerhapsOnlyNumbers.Length > 0;
foreach (char c in lineWithPerhapsOnlyNumbers)
{
if (!char.IsDigit(c))
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
{
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
allNumbers = false;
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
break;
}
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
}
if (allNumbers && lineWithPerhapsOnlyNumbers.Length > 5)
{
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
string[] arr = line.Split(splitChars, StringSplitOptions.RemoveEmptyEntries);
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
if (arr.Length == 1)
{
string[] tc = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (p == null || Math.Abs(p.EndTime.TotalMilliseconds) > 0.001)
{
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
}
p = new Paragraph { StartTime = DecodeTime(tc) };
}
else
{
p.EndTime = DecodeTime(tc);
}
}
}
if (p != null && !allNumbers && line.Length > 1)
{
line = line.Trim();
if (line.StartsWith("}{}", StringComparison.Ordinal) || line.StartsWith("][]", StringComparison.Ordinal))
line = line.Remove(0, 3);
sb.AppendLine(line.Trim());
}
}
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
subtitle.Renumber();
return subtitle;
}
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
private static Subtitle ImportTimeCodesAndTextOnSameLine(List<string> lines)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
var regexTimeCodes1 = new Regex(@"\d+[:.,;]{1}\d\d[:.,;]{1}\d\d[:.,;]{1}\d+", RegexOptions.Compiled);
var regexTimeCodes2 = new Regex(@"\d+[:.,;]{1}\d\d[:.,;]{1}\d+", RegexOptions.Compiled);
Paragraph p = null;
var subtitle = new Subtitle();
var sb = new StringBuilder();
bool isFirstLineNumber = false;
int count = -1;
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
for (int idx = 0; idx < lines.Count; idx++)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
string line = lines[idx];
var matches = regexTimeCodes1.Matches(line);
if (matches.Count == 0)
matches = regexTimeCodes2.Matches(line);
if (matches.Count == 2)
{
var start = matches[0].Value.Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
int i;
if (int.TryParse(start[0], out i))
{
if (count == -1 && i < 2)
count = i;
if (count != i)
{
isFirstLineNumber = false;
break;
}
count++;
}
}
if (count > 2)
isFirstLineNumber = true;
}
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
for (int idx = 0; idx < lines.Count; idx++)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
string line = lines[idx];
if (isFirstLineNumber)
{
while (line.Length > 0 && char.IsDigit(line[0]))
{
line = line.Remove(0, 1);
}
}
var matches = regexTimeCodes1.Matches(line);
if (matches.Count == 0)
matches = regexTimeCodes2.Matches(line);
if (matches.Count == 2)
{
string[] start = matches[0].ToString().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
string[] end = matches[1].ToString().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if ((start.Length == 3 || start.Length == 4) && (end.Length == 3 || end.Length == 4))
{
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
p.StartTime = DecodeTime(start);
p.EndTime = DecodeTime(end);
}
if (matches[0].Index < 9)
line = line.Remove(0, matches[0].Index);
line = line.Replace(matches[0].ToString(), string.Empty);
line = line.Replace(matches[1].ToString(), string.Empty);
line = line.Trim();
if (line.StartsWith("}{}", StringComparison.Ordinal) || line.StartsWith("][]", StringComparison.Ordinal))
line = line.Remove(0, 3);
line = line.Trim();
}
if (p != null && line.Length > 1)
sb.AppendLine(line.Trim());
}
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
// remove all equal headers
if (subtitle.Paragraphs.Count > 5)
{
string prefix = subtitle.Paragraphs[0].Text;
foreach (Paragraph paragraph in subtitle.Paragraphs)
{
string text = paragraph.Text.Trim();
var newPrefix = new StringBuilder();
int i = 0;
while (i < prefix.Length && i < text.Length && text[i] == prefix[i])
{
newPrefix.Append(text[i]);
i++;
}
prefix = newPrefix.ToString();
}
if (prefix.Length > 3 && prefix[1] == ':' && prefix[2] == '\\')
prefix = string.Empty;
if (prefix.Length > 0)
{
foreach (Paragraph paragraph in subtitle.Paragraphs)
{
string text = paragraph.Text.Trim();
Minor refact
2017-04-17 11:15:57 +02:00
if (text.StartsWith(prefix, StringComparison.Ordinal))
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
paragraph.Text = text.Remove(0, prefix.Length);
}
}
}
subtitle.Renumber();
return subtitle;
}
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
private static Subtitle ImportTimeCodesAndTextOnSameLineOnlySpaceAsSeparator(List<string> lines)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
var regexTimeCodes1 = new Regex(@"\d+ {1}\d\d {1}\d\d {1}\d+", RegexOptions.Compiled);
var regexTimeCodes2 = new Regex(@"\d+ {1}\d\d {1}\d+", RegexOptions.Compiled);
Paragraph p = null;
var subtitle = new Subtitle();
var sb = new StringBuilder();
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
char[] splitChar = { ' ' };
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
for (int idx = 0; idx < lines.Count; idx++)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
string line = lines[idx];
var matches = regexTimeCodes1.Matches(line);
if (matches.Count == 0)
matches = regexTimeCodes2.Matches(line);
if (matches.Count == 2)
{
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
string[] start = matches[0].ToString().Split(splitChar, StringSplitOptions.RemoveEmptyEntries);
string[] end = matches[1].ToString().Split(splitChar, StringSplitOptions.RemoveEmptyEntries);
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
if ((start.Length == 3 || start.Length == 4) && (end.Length == 3 || end.Length == 4))
{
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
p.StartTime = DecodeTime(start);
p.EndTime = DecodeTime(end);
}
if (matches[0].Index < 9)
line = line.Remove(0, matches[0].Index);
line = line.Replace(matches[0].ToString(), string.Empty);
line = line.Replace(matches[1].ToString(), string.Empty);
line = line.Trim();
if (line.StartsWith("}{}", StringComparison.Ordinal) || line.StartsWith("][]", StringComparison.Ordinal))
line = line.Remove(0, 3);
line = line.Trim();
}
if (p != null && line.Length > 1)
sb.AppendLine(line.Trim());
}
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
subtitle.Renumber();
return subtitle;
}
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
private static Subtitle ImportTimeCodesOnSameSeperateLine(List<string> lines)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
Paragraph p = null;
var subtitle = new Subtitle();
var sb = new StringBuilder();
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
char[] splitChars = { ' ', '\t' };
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
for (int idx = 0; idx < lines.Count; idx++)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
string line = lines[idx];
string lineWithPerhapsOnlyNumbers = GetLineWithPerhapsOnlyNumbers(line);
bool allNumbers = lineWithPerhapsOnlyNumbers.Length > 0;
foreach (char c in lineWithPerhapsOnlyNumbers)
{
if (!char.IsDigit(c))
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
{
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
allNumbers = false;
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
break;
}
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
}
if (allNumbers && lineWithPerhapsOnlyNumbers.Length > 5)
{
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
string[] arr = line.Replace('-', ' ').Replace('>', ' ').Replace('{', ' ').Replace('}', ' ').Replace('[', ' ').Replace(']', ' ').Trim().Split(splitChars, StringSplitOptions.RemoveEmptyEntries);
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
if (arr.Length == 2)
{
string[] start = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
string[] end = arr[1].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if ((start.Length == 3 || start.Length == 4) && (end.Length == 3 || end.Length == 4))
{
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
p.StartTime = DecodeTime(start);
p.EndTime = DecodeTime(end);
}
}
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
else if (arr.Length > 3)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
string[] start;
string[] end;
if (arr[0].Length > 9)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
start = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
end = arr[1].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
}
else
{
start = arr[1].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
end = arr[2].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
}
if ((start.Length == 3 || start.Length == 4) && (end.Length == 3 || end.Length == 4))
{
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
sb = new StringBuilder();
p.StartTime = DecodeTime(start);
p.EndTime = DecodeTime(end);
}
}
}
if (p != null && !allNumbers && line.Length > 1)
{
line = line.Trim();
if (line.StartsWith("}{}", StringComparison.Ordinal) || line.StartsWith("][]", StringComparison.Ordinal))
line = line.Remove(0, 3);
sb.AppendLine(line.Trim());
}
}
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
double averateDuration = 0;
foreach (Paragraph a in subtitle.Paragraphs)
{
double d = a.Duration.TotalSeconds;
if (d > 10)
d = 8;
averateDuration += d;
}
averateDuration = averateDuration / subtitle.Paragraphs.Count;
if (averateDuration < 0.2 || (averateDuration < 0.5 && subtitle.Paragraphs.Count > 100 && subtitle.Paragraphs[subtitle.Paragraphs.Count - 1].StartTime.TotalSeconds < 140 && subtitle.Paragraphs[subtitle.Paragraphs.Count - 2].StartTime.TotalSeconds < 140))
{
subtitle = ImportTimeCodesOnSameSeperateLineNoMilliseconds(lines);
int i = 0;
foreach (Paragraph a in subtitle.Paragraphs)
{
i++;
var next = subtitle.GetParagraphOrDefault(i);
if (next != null && a.EndTime.TotalMilliseconds >= next.StartTime.TotalMilliseconds)
{
a.EndTime.TotalMilliseconds = next.StartTime.TotalMilliseconds - Configuration.Settings.General.MinimumMillisecondsBetweenLines;
}
}
return subtitle;
}
subtitle.Renumber();
return subtitle;
}
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
private static Subtitle ImportTimeCodesOnSameSeperateLineNoMilliseconds(List<string> lines)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
Paragraph p = null;
var subtitle = new Subtitle();
var sb = new StringBuilder();
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
char[] splitChar = { ' ' };
Optimize "SplitToLines" method (about 60% faster)
2017-11-26 17:35:00 +01:00
for (int idx = 0; idx < lines.Count; idx++)
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
{
string line = lines[idx];
string lineWithPerhapsOnlyNumbers = GetLineWithPerhapsOnlyNumbers(line);
bool allNumbers = lineWithPerhapsOnlyNumbers.Length > 0;
foreach (char c in lineWithPerhapsOnlyNumbers)
{
if (!char.IsDigit(c))
allNumbers = false;
}
if (allNumbers && lineWithPerhapsOnlyNumbers.Length > 5)
{
Updated generic subtitle parser to handle time code lines starting with line# - thx Hanno :)
2017-03-12 15:46:43 +01:00
string[] arr = line.Replace('-', ' ').Replace('>', ' ').Replace('{', ' ').Replace('}', ' ').Replace('[', ' ').Replace(']', ' ').Trim().Split(splitChar, StringSplitOptions.RemoveEmptyEntries);
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
if (arr.Length == 2)
{
string[] start = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
string[] end = arr[1].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if ((start.Length == 3 || start.Length == 4) && (end.Length == 3 || end.Length == 4))
{
if (start.Length == 3)
start = (arr[0].Trim() + ".000").Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (end.Length == 3)
end = (arr[1].Trim() + ".000").Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
p.StartTime = DecodeTime(start);
p.EndTime = DecodeTime(end);
}
}
else if (arr.Length == 3)
{
string[] start = arr[0].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
string[] end = arr[1].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
string[] duration = arr[2].Trim().Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (start.Length == 3)
start = (arr[0].Trim() + ".000").Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (end.Length == 3)
end = (arr[1].Trim() + ".000").Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (duration.Length == 3)
duration = (arr[2].Trim() + ".000").Split(ExpectedSplitChars, StringSplitOptions.RemoveEmptyEntries);
if (start.Length < 3)
{
start = end;
end = duration;
}
if ((start.Length == 3 || start.Length == 4) && (end.Length == 3 || end.Length == 4))
{
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
p = new Paragraph();
Minor perf fix
2017-01-20 22:25:06 +01:00
sb.Clear();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
p.StartTime = DecodeTime(start);
p.EndTime = DecodeTime(end);
}
}
}
if (p != null && !allNumbers && line.Length > 1)
{
line = line.Trim();
if (line.StartsWith("}{}", StringComparison.Ordinal) || line.StartsWith("][]", StringComparison.Ordinal))
line = line.Remove(0, 3);
sb.AppendLine(line.Trim());
}
}
if (p != null)
{
p.Text = sb.ToString().Trim();
subtitle.Paragraphs.Add(p);
}
subtitle.Renumber();
return subtitle;
}
private static string GetLineWithPerhapsOnlyNumbers(string line)
{
Optimize two string operations
2017-12-10 22:18:13 +01:00
return line.RemoveChar(' ').RemoveChar('.').RemoveChar(',').RemoveChar('\t').RemoveChar(':').RemoveChar(';').RemoveChar('{').RemoveChar('}').RemoveChar('[').RemoveChar(']').RemoveChar('-').RemoveChar('>').RemoveChar('<');
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
}
private static TimeCode DecodeTime(string[] parts)
{
try
{
string hour = parts[0];
string minutes = parts[1];
string seconds = parts[2];
string frames;
if (parts.Length < 4)
{
frames = seconds;
seconds = minutes;
minutes = hour;
hour = "0";
}
else
{
frames = parts[3];
}
if (frames.Length < 3)
return new TimeCode(int.Parse(hour), int.Parse(minutes), int.Parse(seconds), SubtitleFormat.FramesToMillisecondsMax999(int.Parse(frames)));
return new TimeCode(int.Parse(hour), int.Parse(minutes), int.Parse(seconds), int.Parse(frames));
}
catch
{
[TimeCode] - Add default constructor. The new default constructor will return new instance of TimeCode with all properties zeroed-out.
2016-10-29 16:56:39 +02:00
return new TimeCode();
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
}
}
Allow import of SRT without line breaks
2017-11-19 13:32:18 +01:00
private static Subtitle ImportSubtitleWithNoLineBreaks(string text)
{
var regex = new Regex(@"^\d+ \d+:\d+:\d+[.,:;]\d+ --> \d+:\d+:\d+[.,:;]\d+\b", RegexOptions.Compiled); // e.g.: 1 00:00:01.502 --> 00:00:03.604
var subtitle = new Subtitle();
int i = 0;
var sb = new StringBuilder();
Paragraph p = null;
while (i < text.Length)
{
var ch = text[i];
if (char.IsNumber(ch))
{
var macth = regex.Match(text.Substring(i));
if (macth.Success)
{
if (p != null)
p.Text = Utilities.AutoBreakLine(sb.ToString().Trim());
sb.Clear();
var arr = macth.Value.Split(' ');
if (arr.Length == 4)
{
i += macth.Value.Length;
p = new Paragraph();
p.StartTime = DecodeTime(arr[1].Split(ExpectedSplitChars));
p.EndTime = DecodeTime(arr[3].Split(ExpectedSplitChars));
subtitle.Paragraphs.Add(p);
continue;
}
}
}
sb.Append(ch);
i++;
}
if (p != null && string.IsNullOrEmpty(p.Text))
p.Text = Utilities.AutoBreakLine(sb.ToString().Trim());
subtitle.Renumber();
return subtitle;
}
Some line endings to CRLF
2016-02-08 21:11:03 +01:00
}
}
Reference in New Issue Copy Permalink