Fix for Whisper Post-processing - thx Purfview/cvrle77 :)

Somewhat related to  #8044

src/Test/Core/AudioToTextTest.cs

@@ -115,5 +115,107 @@ Lexington, Massachusetts.";
             Assert.IsTrue(fixedSubtitle.Paragraphs[4].Text.EndsWith("camp.", StringComparison.Ordinal));
             Assert.IsTrue(fixedSubtitle.Paragraphs[5].Text.StartsWith("Hey there", StringComparison.Ordinal));
         }
+
+        [TestMethod]
+        public void TryForWholeSentences1()
+        {
+            var raw = @"12
+00:00:25,500 --> 00:00:27,060
+Oh, my... Bob, right?
+
+13
+00:00:28,560 --> 00:00:29,220
+Could be fun.
+
+14
+00:00:29,660 --> 00:00:32,580
+Well, we could get to know each other a
+little, maybe loosen things up around
+
+15
+00:00:32,580 --> 00:00:33,060
+here?
+
+16
+00:00:33,680 --> 00:00:39,160
+Well, I've worked with this lot before,
+and, erm... Yeah, this is as loose as they
+
+17
+00:00:39,160 --> 00:00:40,300
+get.
+
+18
+00:00:46,120 --> 00:00:46,340
+Hmm.
+
+19
+00:00:48,160 --> 00:00:49,120
+What's the about that, Bob's?
+
+20
+00:00:49,120 --> 00:00:49,860
+Oh, no.
+
+21
+00:00:50,580 --> 00:00:50,700
+Yep.
+
+22
+00:00:51,240 --> 00:00:52,600
+I felt that soon as I said it.
+
+23
+00:00:54,860 --> 00:00:56,340
+Right, I'm headed out.
+
+24
+00:00:57,460 --> 00:00:58,780
+Everyone have a great day, yeah?
+
+25
+00:00:58,780 --> 00:00:59,600
+Yeah.
+
+26
+00:01:00,600 --> 00:01:01,880
+Wait.
+
+27
+00:01:01,880 --> 00:01:02,480
+Wait.";
+
+            var subtitle = new Subtitle();
+            new SubRip().LoadSubtitle(subtitle, raw.SplitToLines(), null);
+
+            var fixedSubtitle = AudioToTextPostProcessor.TryForWholeSentences(subtitle, "en", 42);
+
+            Assert.AreEqual(14, fixedSubtitle.Paragraphs.Count);
+            Assert.IsTrue(fixedSubtitle.Paragraphs[2].Text == "Well, we could get to know each other a" + Environment.NewLine + "little, maybe loosen things up around here?");
+            Assert.IsTrue(fixedSubtitle.Paragraphs[3].Text == "Well, I've worked with this lot before, and," + Environment.NewLine + "erm... Yeah, this is as loose as they get.");
+            Assert.IsTrue(fixedSubtitle.Paragraphs[4].Text == "Hmm.");
+            Assert.IsTrue(fixedSubtitle.Paragraphs[5].Text == "What's the about that, Bob's?");
+        }
+
+        [TestMethod]
+        public void TryForWholeSentences2()
+        {
+            var raw = @"1
+00:00:26,500 --> 00:00:27,060
+Yes, I think this could indeed be very good. But also
+
+2
+00:00:28,560 --> 00:00:29,220
+that could be fun indeed my friend.";
+
+            var subtitle = new Subtitle();
+            new SubRip().LoadSubtitle(subtitle, raw.SplitToLines(), null);
+
+            var fixedSubtitle = AudioToTextPostProcessor.TryForWholeSentences(subtitle, "en", 42);
+
+            Assert.AreEqual(2, fixedSubtitle.Paragraphs.Count);
+            Assert.IsTrue(fixedSubtitle.Paragraphs[0].Text == "Yes, I think this could" + Environment.NewLine + "indeed be very good.");
+            Assert.IsTrue(fixedSubtitle.Paragraphs[1].Text == "But also that could be" + Environment.NewLine + "fun indeed my friend.");
+        }
     }
 }

src/Test/Logic/AutoTranslate/MergeAndSplitHelperTest.cs

@@ -172,15 +172,22 @@ namespace Test.Logic.AutoTranslate
             Assert.IsTrue(splitResult[0].Length > 5);
             Assert.IsTrue(splitResult[1].Length > 5);
             Assert.IsTrue(splitResult[2].Length > 5);
-            Assert.AreEqual(string.Join(" ", subtitle.Paragraphs.Select(p => p.Text)), string.Join(" ", splitResult));
+
+            var subtitleText = string.Join("", subtitle.Paragraphs.Select(p => p.Text)).RemoveChar('\n', '\r', ' ');
+            var splitText = string.Join("", splitResult).RemoveChar('\n', '\r', ' ');
+            Assert.AreEqual(subtitleText, splitText);
+
+            Assert.AreEqual("Hallo there. In the" + Environment.NewLine + "garden today are we?", splitResult[0]);
+            Assert.AreEqual("So I will very", splitResult[1]);
+            Assert.AreEqual("soon be going home to Sweden.", splitResult[2]);
             Assert.AreEqual("My name is Peter!", splitResult[3]);
             Assert.AreEqual("My name is Peter! And Jones.", splitResult[4]);
             Assert.AreEqual("My name is Peter. And Jones.", splitResult[5]);
             Assert.AreEqual("", splitResult[6]);
             Assert.AreEqual("Hallo there.", splitResult[7]);
 
-            var inputText = string.Join(" ", subtitle.Paragraphs.Select(p => p.Text)).Replace(Environment.NewLine, " ");
+            var inputText = string.Join("", subtitle.Paragraphs.Select(p => p.Text)).RemoveChar('\n', '\r', ' ');
-            var splitResultText = string.Join(" ", splitResult);
+            var splitResultText = string.Join("", splitResult).RemoveChar('\n', '\r', ' ');
             Assert.AreEqual(inputText, splitResultText);
         }
     }

src/libse/AudioToText/AudioToTextPostProcessor.cs

@@ -64,11 +64,7 @@ namespace Nikse.SubtitleEdit.Core.AudioToText
                         continue;
                     }
 
-                    if (TwoLetterLanguageCode == "en")
+                    if (TwoLetterLanguageCode == "da")
-                    {
-                        // anything?
-                    }
-                    else if (TwoLetterLanguageCode == "da")
                     {
                         if (paragraph.Text.Contains("Danske tekster af nicolai winther", StringComparison.OrdinalIgnoreCase))
                         {
@@ -144,10 +140,12 @@ namespace Nikse.SubtitleEdit.Core.AudioToText
             return subtitle;
         }
 
-        public static Subtitle TryForWholeSentences(Subtitle inputSubtitle, string language, int lineMxLength)
+        public static Subtitle TryForWholeSentences(Subtitle inputSubtitle, string language, int lineMaxLength)
         {
             var s = new Subtitle(inputSubtitle);
             const int maxMoveChunkSize = 15;
+            var deleteIndices = new List<int>();
+
             for (var i = 0; i < s.Paragraphs.Count - 1; i++)
             {
                 var p = s.Paragraphs[i];
@@ -158,14 +156,21 @@ namespace Nikse.SubtitleEdit.Core.AudioToText
                     p.EndTime.TotalMilliseconds - next.StartTime.TotalMilliseconds > 100 ||
                     p.Text.Contains('<') ||
                     next.Text.Contains('<') ||
-                    !(p.Text.Contains('.') || next.Text.Contains('.')) ||
+                    !(p.Text.Contains('.') || p.Text.Contains('?') || p.Text.Contains('!') || next.Text.Contains('.') || next.Text.Contains('?') || next.Text.Contains('!')) ||
-                    p.Text.EndsWith('.'))
+                    p.Text.EndsWith('.') ||
+                    p.Text.EndsWith('?') ||
+                    p.Text.EndsWith('!'))
+                {
+                    continue;
+                }
+
+                if (deleteIndices.Contains(i))
                 {
                     continue;
                 }
 
                 // check for period in last part of p
-                var lastPeriodIdx = p.Text.LastIndexOf('.');
+                var lastPeriodIdx = p.Text.LastIndexOfAny(new char[] { '.', '?', '!' });
                 if (lastPeriodIdx > 3 && lastPeriodIdx > p.Text.Length - maxMoveChunkSize)
                 {
                     var newCurrentText = p.Text.Substring(0, lastPeriodIdx + 1).Trim();
@@ -177,23 +182,31 @@ namespace Nikse.SubtitleEdit.Core.AudioToText
                     var arrayCurrent = newCurrentText.SplitToLines();
                     var arrayNext = newNextText.SplitToLines();
 
-                    var currentOk = arrayCurrent.Count == 1 || (arrayCurrent.Count == 2 && arrayCurrent[0].Length <= lineMxLength);
+                    var currentOk = arrayCurrent.Count == 1 || (arrayCurrent.Count == 2 && arrayCurrent[0].Length < lineMaxLength * 2);
-                    var nextOk = arrayNext.Count == 1 || (arrayNext.Count == 2 && arrayNext[0].Length <= lineMxLength);
+                    var nextOk = arrayNext.Count == 1 || (arrayNext.Count == 2 && arrayNext[0].Length < lineMaxLength * 2);
 
                     if (currentOk && nextOk)
                     {
                         p.Text = newCurrentText;
                         next.Text = newNextText;
 
-                        //TODO: calc time
+                        if (string.IsNullOrWhiteSpace(newCurrentText))
+                        {
+                            deleteIndices.Add(i);
+                            next.StartTime.TotalMilliseconds = p.StartTime.TotalMilliseconds;
+                        }
+                        else
+                        {
+                            //TODO: calc time
+                        }
 
                         continue;
                     }
                 }
 
                 // check for period in beginning of next
-                var firstPeriodIdx = next.Text.IndexOf('.');
+                var firstPeriodIdx = next.Text.IndexOfAny(new char[] { '.', '?', '!' });
-                if (firstPeriodIdx > 3 && firstPeriodIdx < maxMoveChunkSize)
+                if (firstPeriodIdx >= 3 && firstPeriodIdx < maxMoveChunkSize)
                 {
                     var newCurrentText = next.Text.Substring(0, firstPeriodIdx + 1).Trim();
                     var newNextText = next.Text.Remove(0, firstPeriodIdx + 1).Trim();
@@ -204,19 +217,29 @@ namespace Nikse.SubtitleEdit.Core.AudioToText
                     var arrayCurrent = newCurrentText.SplitToLines();
                     var arrayNext = newNextText.SplitToLines();
 
-                    var currentOk = arrayCurrent.Count == 1 || (arrayCurrent.Count == 2 && arrayCurrent[0].Length <= lineMxLength);
+                    var currentOk = arrayCurrent.Count == 1 || (arrayCurrent.Count == 2 && arrayCurrent[0].Length < lineMaxLength * 2);
-                    var nextOk = arrayNext.Count == 1 || (arrayNext.Count == 2 && arrayNext[0].Length <= lineMxLength);
+                    var nextOk = arrayNext.Count == 1 || (arrayNext.Count == 2 && arrayNext[0].Length < lineMaxLength * 2);
 
                     if (currentOk && nextOk)
                     {
                         p.Text = newCurrentText;
                         next.Text = newNextText;
 
-                        //TODO: calc time
+                        if (string.IsNullOrWhiteSpace(newNextText))
+                        {
+                            deleteIndices.Add(i + 1);
+                            p.EndTime.TotalMilliseconds = next.EndTime.TotalMilliseconds;
+                        }
+                        else
+                        {
+                            //TODO: calc time
+                        }
                     }
                 }
             }
 
+            s.RemoveParagraphsByIndices(deleteIndices);
+
             return s;
         }
 

src/libse/Common/Subtitle.cs

@@ -756,7 +756,7 @@ namespace Nikse.SubtitleEdit.Core.Common
         /// <returns>Number of lines deleted</returns>
         public int RemoveParagraphsByIndices(IEnumerable<int> indices)
         {
-            int count = 0;
+            var count = 0;
             foreach (var index in indices.OrderByDescending(p => p))
             {
                 if (index >= 0 && index < Paragraphs.Count)
@@ -765,6 +765,7 @@ namespace Nikse.SubtitleEdit.Core.Common
                     count++;
                 }
             }
+
             return count;
         }