Subtitles - algorithm

[Delta2]

Hi,

I think everyone knows Subtitle Workshop, if not, well there is a lot of subtitles tools around

Well, I made a DOT.NET APP mostly for my personal use, that accomplish AVI->DVD conversion, MPG-1 -> DVD, MPG-2 -> DVD, and burning itself a DVD movie at final stage. Works plenty well

I want to join subtitles and recalculate time values, but this is a little hard because subtitles files are TEXT FILES, nothing indexed for direct reading, and I must read each line, etc...

Where can I get a Subtitles algorithm mainly for joining. Or does someone hava an idea how to do this easily ?

[Delta2]

never mind

I made my one code

usage:

(legendas=subtitles)
SubFiles is a string array with subtitles full path, like :

SubFiles(0)="c:\dvd\subtitle1.srt"
SubFiles(2)="c:\dvd\subtitle2.srt"

Dim legendas As New SubsClass(SubFiles, WorkingPath, Me)
legendas.Join()



=====================
Imports System.Runtime.InteropServices
Imports System
Imports System.Drawing
Imports System.Windows.Forms
Imports System.IO
Imports Microsoft.Win32
Imports System.Diagnostics
Imports System.ComponentModel


Public Class SubsClass
Private formObj As Form1
Private Structure legenda
Dim old_text As String()
Dim new_text As String()
Dim old_timings() As String
Dim new_timings() As String
End Structure

Private SubFiles(20) As legenda
Private Files() As String
Private WorkingPath As String


Public Sub Join()
Dim i, j As Integer
Dim startTime(3) As Integer
Dim endTime(3) As Integer
Dim tempos() As String
Dim ultimoTempo As TimeSpan

'
Dim inicio As TimeSpan
Dim fim As TimeSpan
Dim oldTiming As String
Dim newTiming As String

'hours + minutes + seconds + milliseconds

'00:00:00,030 --> 00:00:20,470

'
'tempos(0) - início
'tempos(1) - fim
'
i = UBound(SubFiles(0).old_timings)
tempos = Strings.Split(SubFiles(0).old_timings(i), "-->")
tempos(0) = Trim(tempos(0))
tempos(1) = Trim(tempos(1))

endTime(0) = CInt(Strings.Split(tempos(1), ":")(0))
endTime(1) = CInt(Strings.Split(tempos(1), ":")(1))
endTime(2) = CInt(Strings.Left(Strings.Split(tempos(1), ":")(2), 2))
endTime(3) = CInt(Strings.Right(Strings.Split(tempos(1), ":")(2), 3))

ultimoTempo = New TimeSpan(0, endTime(0), endTime(1), endTime(2), endTime(3))

For i = 0 To UBound(SubFiles(1).old_timings)
tempos = Strings.Split(SubFiles(1).old_timings(i), "-->")
tempos(0) = Trim(tempos(0))
tempos(1) = Trim(tempos(1))

startTime(0) = CInt(Strings.Split(tempos(0), ":")(0))
startTime(1) = CInt(Strings.Split(tempos(0), ":")(1))
startTime(2) = CInt(Strings.Left(Strings.Split(tempos(0), ":")(2), 2))
startTime(3) = CInt(Strings.Right(Strings.Split(tempos(0), ":")(2), 3))

endTime(0) = CInt(Strings.Split(tempos(1), ":")(0))
endTime(1) = CInt(Strings.Split(tempos(1), ":")(1))
endTime(2) = CInt(Strings.Left(Strings.Split(tempos(1), ":")(2), 2))
endTime(3) = CInt(Strings.Right(Strings.Split(tempos(1), ":")(2), 3))

inicio = New TimeSpan(0, startTime(0), startTime(1), startTime(2), startTime(3))
fim = New TimeSpan(0, endTime(0), endTime(1), endTime(2), endTime(3))

inicio = inicio.Add(ultimoTempo)
fim = fim.Add(ultimoTempo)

newTiming = CStr(inicio.Hours) & ":" & CStr(inicio.Minutes) & ":" & CStr(inicio.Seconds) & "," & CStr(inicio.Milliseconds)
newTiming = newTiming & " --> "
newTiming = newTiming & CStr(fim.Hours) & ":" & CStr(fim.Minutes) & ":" & CStr(fim.Seconds) & "," & CStr(fim.Milliseconds)

SubFiles(1).new_timings(i) = newTiming
Next

Dim contador As Integer = 0
For i = 0 To UBound(SubFiles(1).old_text)

If (InStr(SubFiles(1).old_text(i), "-->") > 0) And (InStr(SubFiles(1).old_text(i), ":") > 0) Then
SubFiles(1).new_text(i) = SubFiles(1).new_timings(contador)
contador = contador + 1
Else
SubFiles(1).new_text(i) = SubFiles(1).old_text(i)
End If
Next

Dim ultimaLegenda As Integer

Dim A, B As String
For i = 0 To UBound(SubFiles(0).old_text)
If IsNumeric(Trim(SubFiles(0).old_text(i))) Then ultimaLegenda = CInt(Trim(SubFiles(0).old_text(i)))
A = A + SubFiles(0).old_text(i) & vbCrLf
Next

contador = ultimaLegenda + 1

For i = 0 To UBound(SubFiles(1).new_text)
If IsNumeric(Trim(SubFiles(1).new_text(i))) Then
SubFiles(1).new_text(i) = CStr(contador)
contador = contador + 1
End If
B = B + SubFiles(1).new_text(i) & vbCrLf
Next

Dim outfile As New StreamWriter(WorkingPath & "legendas.srt", False, System.Text.Encoding.GetEncoding("iso-8859-1"))
Try
outfile.Write(A + B)
Finally
outfile.Close()
End Try
End Sub

Public Sub loadLegendas()
Dim sr As StreamReader
Dim contador, i, j As Integer


For i = 0 To UBound(Files)
sr = New StreamReader(Files(i), System.Text.Encoding.GetEncoding("iso-8859-1"))
SubFiles(i).old_text = Split(sr.ReadToEnd, vbCrLf)
sr = Nothing

SubFiles(i).new_text = SubFiles(i).old_text

contador = 0
ReDim SubFiles(i).old_timings(5000)
ReDim SubFiles(i).new_timings(5000)

For j = 0 To UBound(SubFiles(i).old_text)
If (InStr(SubFiles(i).old_text(j), "-->") > 0) And (InStr(SubFiles(i).old_text(j), ":") > 0) Then
SubFiles(i).old_timings(contador) = SubFiles(i).old_text(j)
SubFiles(i).new_timings(contador) = SubFiles(i).old_text(j)
contador = contador + 1
End If
Next

ReDim Preserve SubFiles(i).old_timings(contador - 1)
ReDim Preserve SubFiles(i).new_timings(contador - 1)
Next
End Sub

Public Sub New(ByRef subs() As String, ByVal path As String, ByRef frm As Form1)
Files = subs
ReDim SubFiles(UBound(Files))
WorkingPath = path
formObj = frm
loadLegendas()
End Sub
End Class

[slyzaar]

Hi Delta2,

I am playing around with a few video capture cards which encode to MPEG2.

One of my requirements is to also extract/capture the subtitle data.

WHat I have found is that the capture cards seem to output the subtitles as bitmaps and not as strings - have you come across this?

I would be eternally greatfull if you could shed some light on how I can extract the subtitles as string.

Many thanks,

KS

[borax]

If the subs are bitmaps, there is no other way but OCR to get strings. Or use the bitmaps directly if the 'target' is a DVD. On DVDs the subs are bitmaps encoded in a 2Bit RLE format.

If the subs are part of a DVB straem, these could be found as strings in the teletext. If that's the case, one of the best programs to look at is projectx.