Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-23 03:33:18 +01:00
Code Issues Projects Releases Wiki Activity
f552af4555
SubtitleEdit/libse/SubtitleFormats/Sami.cs

441 lines
18 KiB
C#
Raw Blame History

using System;
using System.Collections.Generic;
using System.Globalization;
using System.Net;
using System.Text;
using System.Text.RegularExpressions;
namespace Nikse.SubtitleEdit.Core.SubtitleFormats
{
public class Sami : SubtitleFormat
{
public override string Extension
{
get { return ".smi"; }
}
public override string Name
{
get { return "SAMI"; }
}
public override bool IsTimeBased
{
get { return true; }
}
public override bool IsMine(List<string> lines, string fileName)
{
var sb = new StringBuilder();
foreach (string l in lines)
sb.AppendLine(l);
if (sb.ToString().Contains("</SYNC>"))
return false;
var subtitle = new Subtitle();
LoadSubtitle(subtitle, lines, fileName);
return subtitle.Paragraphs.Count > _errorCount;
}
public override string ToText(Subtitle subtitle, string title)
{
string language = Utilities.AutoDetectLanguageName("en_US", subtitle);
var ci = CultureInfo.GetCultureInfo(language.Replace("_", "-"));
string languageTag = string.Format("{0}CC", language.Replace("_", string.Empty).ToUpper());
string languageName = ci.Parent.EnglishName;
string languageStyle = string.Format(".{0} [ name: {1}; lang: {2} ; SAMIType: CC ; ]", languageTag, languageName, language.Replace("_", "-"));
languageStyle = languageStyle.Replace("[", "{").Replace("]", "}");
string header =
@"<SAMI>
<HEAD>
<TITLE>_TITLE_</TITLE>
<SAMIParam>
Metrics {time:ms;}
Spec {MSFT:1.0;}
</SAMIParam>
<STYLE TYPE=""text/css"">
<!--
P { font-family: Arial; font-weight: normal; color: white; background-color: black; text-align: center; }
_LANGUAGE-STYLE_
-->
</STYLE>
</HEAD>
<BODY>
<-- Open play menu, choose Captions and Subtiles, On if available -->
<-- Open tools menu, Security, Show local captions when present -->";
bool useExtra = false;
if (!string.IsNullOrEmpty(subtitle.Header) && subtitle.Header.StartsWith("<style", StringComparison.OrdinalIgnoreCase))
{
useExtra = true;
header =
@"<SAMI>
<HEAD>
<TITLE>_TITLE_</TITLE>
<SAMIParam>
Metrics {time:ms;}
Spec {MSFT:1.0;}
</SAMIParam>
" + subtitle.Header.Trim() + @"
</HEAD>
<BODY>
<-- Open play menu, choose Captions and Subtiles, On if available -->
<-- Open tools menu, Security, Show local captions when present -->";
}
// Example text (start numbers are milliseconds)
//<SYNC Start=65264><P>Let's go!
//<SYNC Start=66697><P><BR>
string paragraphWriteFormat = @"<SYNC Start={0}><P Class={3}>{2}" + Environment.NewLine +
@"<SYNC Start={1}><P Class={3}>&nbsp;";
string paragraphWriteFormatOpen = @"<SYNC Start={0}><P Class={2}>{1}";
if (Name == new SamiModern().Name)
{
paragraphWriteFormat = "<SYNC Start=\"{0}\"><P Class=\"{3}\">{2}</P></SYNC>" + Environment.NewLine +
"<SYNC Start=\"{1}\"><P Class=\"{3}\">&nbsp;</P></SYNC>";
paragraphWriteFormatOpen = "<SYNC Start=\"{0}\"><P Class=\"{2}\">{1}</P></SYNC>";
}
else if (Name == new SamiYouTube().Name)
{
paragraphWriteFormat = "<SYNC Start=\"{0}\"><P Class=\"{3}\">{2}</P></SYNC>" + Environment.NewLine +
"<SYNC Start=\"{1}\"><P Class=\"{3}\"></P></SYNC>";
paragraphWriteFormatOpen = "<SYNC Start=\"{0}\"><P Class=\"{2}\">{1}</P></SYNC>";
}
int count = 1;
var sb = new StringBuilder();
sb.AppendLine(header.Replace("_TITLE_", title).Replace("_LANGUAGE-STYLE_", languageStyle));
var totalLine = new StringBuilder();
var partialLine = new StringBuilder();
foreach (Paragraph p in subtitle.Paragraphs)
{
Paragraph next = subtitle.GetParagraphOrDefault(count);
string text = p.Text;
if (text.Contains('<') && text.Contains('>'))
{
bool tagOn = false;
for (int i = 0; i < text.Length; i++)
{
string t = text.Substring(i);
if (t.StartsWith('<') &&
(t.StartsWith("<font", StringComparison.Ordinal) ||
t.StartsWith("<div", StringComparison.Ordinal) ||
t.StartsWith("<i", StringComparison.Ordinal) ||
t.StartsWith("<b", StringComparison.Ordinal) ||
t.StartsWith("<s", StringComparison.Ordinal) ||
t.StartsWith("</", StringComparison.Ordinal)))
{
totalLine.Append(EncodeText(partialLine.ToString()));
partialLine.Clear();
tagOn = true;
totalLine.Append('<');
}
else if (t.StartsWith('>') && tagOn)
{
tagOn = false;
totalLine.Append('>');
}
else if (!tagOn)
{
partialLine.Append(text[i]);
}
else
{
totalLine.Append(text[i]);
}
}
totalLine.Append(EncodeText(partialLine.ToString()));
text = totalLine.ToString();
totalLine.Clear();
partialLine.Clear();
}
else
{
text = EncodeText(text);
}
if (Name == new SamiModern().Name)
text = text.Replace(Environment.NewLine, "<br />");
else
text = text.Replace(Environment.NewLine, "<br>");
string currentClass = languageTag;
if (useExtra && !string.IsNullOrEmpty(p.Extra))
currentClass = p.Extra;
var startMs = (long)(Math.Round(p.StartTime.TotalMilliseconds));
var endMs = (long)(Math.Round(p.EndTime.TotalMilliseconds));
if (next != null && Math.Abs(((long)Math.Round(next.StartTime.TotalMilliseconds)) - endMs) < 1)
sb.AppendLine(string.Format(paragraphWriteFormatOpen, startMs, text, currentClass));
else
sb.AppendLine(string.Format(paragraphWriteFormat, startMs, endMs, text, currentClass));
count++;
}
sb.AppendLine("</BODY>");
sb.AppendLine("</SAMI>");
return sb.ToString().Trim();
}
private static string EncodeText(string text)
{
switch (Configuration.Settings.SubtitleSettings.SamiHtmlEncodeMode)
{
case 1:
return WebUtility.HtmlEncode(text);
case 2:
return HtmlUtil.EncodeNamed(text);
case 3:
return HtmlUtil.EncodeNumeric(text);
}
return text;
}
public static List<string> GetStylesFromHeader(string header)
{
var list = new List<string>();
if (!string.IsNullOrEmpty(header) && header.StartsWith("<style", StringComparison.OrdinalIgnoreCase))
{
foreach (string line in header.SplitToLines())
{
string s = line.Trim();
if (s.StartsWith('.') && s.IndexOf(' ') > 2)
{
string name = s.Substring(1, s.IndexOf(' ') - 1);
list.Add(name);
}
}
}
else
{
list.Add("ENUSCC");
}
return list;
}
public override void LoadSubtitle(Subtitle subtitle, List<string> lines, string fileName)
{
_errorCount = 0;
var sb = new StringBuilder();
foreach (string l in lines)
sb.AppendLine(l.Replace("<SYNC Start= \"", "<SYNC Start=\"").Replace("<SYNC Start = \"", "<SYNC Start=\"").Replace("<SYNC Start =\"", "<SYNC Start=\"").Replace("<SYNC Start=\"", "<SYNC Start=\""));
string allInput = sb.ToString();
string allInputLower = allInput.ToLower();
if (!allInputLower.Contains("<sync "))
return;
int styleStart = allInputLower.IndexOf("<style", StringComparison.Ordinal);
if (styleStart > 0)
{
int styleEnd = allInputLower.IndexOf("</style>", StringComparison.Ordinal);
if (styleEnd > 0)
{
subtitle.Header = allInput.Substring(styleStart, styleEnd - styleStart + 8);
}
}
const string syncTag = "<sync start=";
const string syncTagEnc = "<sync encrypted=\"true\" start=";
int syncStartPos = allInputLower.IndexOf(syncTag, StringComparison.Ordinal);
int index = syncStartPos + syncTag.Length;
int syncStartPosEnc = allInputLower.IndexOf(syncTagEnc, StringComparison.Ordinal);
if ((syncStartPosEnc >= 0 && syncStartPosEnc < syncStartPos) || syncStartPos == -1)
{
syncStartPos = syncStartPosEnc;
index = syncStartPosEnc + syncTagEnc.Length;
}
var p = new Paragraph();
while (syncStartPos >= 0)
{
string millisecAsString = string.Empty;
while (index < allInput.Length && @"""'0123456789".Contains(allInput[index]))
{
if (allInput[index] != '"' && allInput[index] != '\'')
millisecAsString += allInput[index];
index++;
}
while (index < allInput.Length && allInput[index] != '>')
index++;
if (index < allInput.Length && allInput[index] == '>')
index++;
int syncEndPos = allInputLower.IndexOf(syncTag, index, StringComparison.Ordinal);
int syncEndPosEnc = allInputLower.IndexOf(syncTagEnc, index, StringComparison.Ordinal);
if ((syncStartPosEnc >= 0 && syncStartPosEnc < syncStartPos) || syncEndPos == -1)
syncEndPos = syncEndPosEnc;
string text;
if (syncEndPos >= 0)
text = allInput.Substring(index, syncEndPos - index);
else
text = allInput.Substring(index);
string textToLower = text.ToLower();
if (textToLower.Contains(" class="))
{
var className = new StringBuilder();
int startClass = textToLower.IndexOf(" class=", StringComparison.Ordinal);
int indexClass = startClass + 7;
while (indexClass < textToLower.Length && (Utilities.LowercaseLettersWithNumbers + @"'""").Contains(textToLower[indexClass]))
{
className.Append(text[indexClass]);
indexClass++;
}
p.Extra = className.ToString().Trim(' ', '\'', '"');
}
if (text.Contains("ID=\"Source\"") || text.Contains("ID=Source"))
{
int sourceIndex = text.IndexOf("ID=\"Source\"", StringComparison.Ordinal);
if (sourceIndex < 0)
sourceIndex = text.IndexOf("ID=Source", StringComparison.Ordinal);
int st = sourceIndex - 1;
while (st > 0 && text.Substring(st, 2).ToUpper() != "<P")
{
st--;
}
if (st > 0)
{
text = text.Substring(0, st) + text.Substring(sourceIndex);
}
int et = st;
while (et < text.Length - 5 && text.Substring(et, 3).ToUpper() != "<P>" && text.Substring(et, 4).ToUpper() != "</P>")
{
et++;
}
text = text.Substring(0, st) + text.Substring(et);
}
text = text.Replace(Environment.NewLine, " ");
text = text.Replace(" ", " ");
text = text.TrimEnd();
text = Regex.Replace(text, @"<br {0,2}/?>", Environment.NewLine, RegexOptions.IgnoreCase);
while (text.Contains(" "))
text = text.Replace(" ", " ");
text = text.Replace("</BODY>", string.Empty).Replace("</SAMI>", string.Empty).TrimEnd();
int endSyncPos = text.ToUpper().IndexOf("</SYNC>", StringComparison.Ordinal);
if (text.IndexOf('>') > 0 && (text.IndexOf('>') < endSyncPos || endSyncPos == -1))
text = text.Remove(0, text.IndexOf('>') + 1);
text = text.TrimEnd();
if (text.EndsWith("</sync>", StringComparison.OrdinalIgnoreCase))
text = text.Substring(0, text.Length - 7).TrimEnd();
if (text.EndsWith("</p>", StringComparison.Ordinal) || text.EndsWith("</P>", StringComparison.Ordinal))
text = text.Substring(0, text.Length - 4).TrimEnd();
text = text.Replace("&nbsp;", " ").Replace("&NBSP;", " ");
if (text.Contains("<font color=") && !text.Contains("</font>"))
text += "</font>";
if (text.StartsWith("<FONT COLOR=") && !text.Contains("</font>") && !text.Contains("</FONT>"))
text += "</FONT>";
if (text.Contains('<') && text.Contains('>'))
{
var total = new StringBuilder();
var partial = new StringBuilder();
bool tagOn = false;
for (int i = 0; i < text.Length && i < 999; i++)
{
string tmp = text.Substring(i);
if (tmp.StartsWith('<') &&
(tmp.StartsWith("<font", StringComparison.Ordinal) ||
tmp.StartsWith("<div", StringComparison.Ordinal) ||
tmp.StartsWith("<i", StringComparison.Ordinal) ||
tmp.StartsWith("<b", StringComparison.Ordinal) ||
tmp.StartsWith("<s", StringComparison.Ordinal) ||
tmp.StartsWith("</", StringComparison.Ordinal)))
{
total.Append(WebUtility.HtmlDecode(partial.ToString()));
partial = new StringBuilder();
tagOn = true;
total.Append('<');
}
else if (text.Substring(i).StartsWith('>') && tagOn)
{
tagOn = false;
total.Append('>');
}
else if (!tagOn)
{
partial.Append(text[i]);
}
else
{
total.Append(text[i]);
}
}
total.Append(WebUtility.HtmlDecode(partial.ToString()));
text = total.ToString();
}
else
{
text = WebUtility.HtmlDecode(text);
}
var cleanText = text.FixExtraSpaces();
cleanText = cleanText.Trim();
if (!string.IsNullOrEmpty(p.Text) && !string.IsNullOrEmpty(millisecAsString))
{
p.EndTime = new TimeCode(long.Parse(millisecAsString));
subtitle.Paragraphs.Add(p);
p = new Paragraph();
}
p.Text = cleanText;
long l;
if (long.TryParse(millisecAsString, out l))
p.StartTime = new TimeCode(l);
if (syncEndPos <= 0)
{
syncStartPos = -1;
}
else
{
syncStartPos = allInputLower.IndexOf(syncTag, syncEndPos, StringComparison.Ordinal);
index = syncStartPos + syncTag.Length;
syncStartPosEnc = allInputLower.IndexOf(syncTagEnc, syncEndPos, StringComparison.Ordinal);
if ((syncStartPosEnc >= 0 && syncStartPosEnc < syncStartPos) || syncStartPos == -1)
{
syncStartPos = syncStartPosEnc;
index = syncStartPosEnc + syncTagEnc.Length;
}
}
}
if (!string.IsNullOrEmpty(p.Text) && !subtitle.Paragraphs.Contains(p))
{
p.EndTime.TotalMilliseconds = p.StartTime.TotalMilliseconds + Utilities.GetOptimalDisplayMilliseconds(p.Text);
subtitle.Paragraphs.Add(p);
}
subtitle.Renumber();
if (subtitle.Paragraphs.Count > 0 &&
(subtitle.Paragraphs[subtitle.Paragraphs.Count - 1].Text.ToUpper().Trim() == "</BODY>" ||
subtitle.Paragraphs[subtitle.Paragraphs.Count - 1].Text.ToUpper().Trim() == "<BODY>"))
subtitle.Paragraphs.RemoveAt(subtitle.Paragraphs.Count - 1);
foreach (Paragraph p2 in subtitle.Paragraphs)
{
p2.Text = p2.Text.Replace('\u00A0', ' '); // non-breaking space to normal space
}
}
public override bool HasStyleSupport
{
get { return true; }
}
}
}
Reference in New Issue View Git Blame Copy Permalink