More work related to word-split-list

Dictionaries/en_US_user.xml

@@ -88,6 +88,7 @@
   <word>colonoscopy</word>
   <word>colours</word>
   <word>contractualism</word>
+  <word>copain</word>
   <word>copernicium</word>
   <word>copper</word>
   <word>copperish</word>
@@ -96,6 +97,9 @@
   <word>cottonoid</word>
   <word>could've</word>
   <word>craniotomy</word>
+  <word>crossbone</word>
+  <word>crosshair</word>
+  <word>crosshairs</word>
   <word>cryo</word>
   <word>cryotube</word>
   <word>cudgelled</word>
@@ -113,6 +117,7 @@
   <word>douchey</word>
   <word>dreamt</word>
   <word>dubnium</word>
+  <word>dumbass</word>
   <word>dysesthesia</word>
   <word>dysprosium</word>
   <word>einsteinium</word>
@@ -130,8 +135,10 @@
   <word>everything's</word>
   <word>extradural</word>
   <word>faggot</word>
+  <word>failsafe</word>
   <word>falafel</word>
   <word>fallin'</word>
+  <word>fanboy</word>
   <word>favour</word>
   <word>favoured</word>
   <word>favourite</word>
@@ -145,6 +152,7 @@
   <word>fluorine</word>
   <word>flushin'</word>
   <word>flyer</word>
+  <word>forevermore</word>
   <word>francium</word>
   <word>fuckable</word>
   <word>fundraiser</word>
@@ -178,6 +186,7 @@
   <word>hematoma</word>
   <word>hiei</word>
   <word>hijab</word>
+  <word>hitman</word>
   <word>hm</word>
   <word>holmium</word>
   <word>hottie</word>
@@ -190,6 +199,7 @@
   <word>hypothermic</word>
   <word>immersive</word>
   <word>immunotherapy</word>
+  <word>inbox</word>
   <word>incontinentia</word>
   <word>indium</word>
   <word>intercostal</word>
@@ -219,6 +229,7 @@
   <word>laryngopharyngeal</word>
   <word>lawrencium</word>
   <word>lead</word>
+  <word>lightsaber</word>
   <word>lithium</word>
   <word>lobectomy</word>
   <word>lockdown</word>
@@ -265,6 +276,8 @@
   <word>neuroscientific</word>
   <word>neurotypical</word>
   <word>neurovascular</word>
+  <word>newsfeed</word>
+  <word>newsfeeds</word>
   <word>nickel</word>
   <word>niobium</word>
   <word>nitrogen</word>
@@ -330,6 +343,8 @@
   <word>rubidium</word>
   <word>ruthenium</word>
   <word>rutherfordium</word>
+  <word>safehouse</word>
+  <word>safeword</word>
   <word>saké</word>
   <word>saltimbocca</word>
   <word>samarium</word>
@@ -346,6 +361,8 @@
   <word>señorita</word>
   <word>sensei</word>
   <word>sharpshoot</word>
+  <word>shitbox</word>
+  <word>shithead</word>
   <word>shithead's</word>
   <word>shithole</word>
   <word>should've</word>
@@ -379,6 +396,7 @@
   <word>sudoku</word>
   <word>sulfur</word>
   <word>sulphur</word>
+  <word>sunglass</word>
   <word>supervolcano</word>
   <word>synchronicity</word>
   <word>syncopal</word>
@@ -440,6 +458,7 @@
   <word>vegetations</word>
   <word>voicemail</word>
   <word>voila</word>
+  <word>walkthrough</word>
   <word>weirding</word>
   <word>what'd</word>
   <word>what're</word>

src/libse/Common/Settings.cs

@@ -135,6 +135,7 @@ namespace Nikse.SubtitleEdit.Core.Common
         public string OcrTrainFonts { get; set; }
         public string OcrTrainMergedLetters { get; set; }
         public string OcrTrainSrtFile { get; set; }
+        public bool OcrUseWordSplitList { get; set; }
         public string BDOpenIn { get; set; }
         public string Interjections { get; set; }
         public string MicrosoftBingApiId { get; set; }
@@ -384,6 +385,7 @@ namespace Nikse.SubtitleEdit.Core.Common
             OcrAddLetterRow2 = "♫;Á;É;Í;Ó;Ö;Ő;Ú;Ü;Ű;Ç;Ñ;Å;¡";
             OcrTrainFonts = "Arial;Calibri;Corbel;Futura Std Book;Futura Bis;Helvetica Neue;Lucida Console;Tahoma;Trebuchet MS;Verdana";
             OcrTrainMergedLetters = "ff ft fi fj fy fl rf rt rv rw ry rt rz ryt tt TV tw yt yw wy wf ryt xy";
+            OcrUseWordSplitList = true;
             Interjections = "Ah;Ahem;Ahh;Ahhh;Ahhhh;Eh;Ehh;Ehhh;Hm;Hmm;Hmmm;Huh;Mm;Mmm;Mmmm;Phew;Gah;Oh;Ohh;Ohhh;Ow;Oww;Owww;Ugh;Ughh;Uh;Uhh;Uhhh;Whew";
             MicrosoftTranslatorTokenEndpoint = "https://api.cognitive.microsoft.com/sts/v1.0/issueToken";
             GoogleTranslateNoKeyWarningShow = true;
@@ -4143,6 +4145,12 @@ $HorzAlign          =   Center
                 settings.Tools.OcrTrainSrtFile = subNode.InnerText;
             }
 
+            subNode = node.SelectSingleNode("OcrUseWordSplitList");
+            if (subNode != null)
+            {
+                settings.Tools.OcrUseWordSplitList = Convert.ToBoolean(subNode.InnerText, CultureInfo.InvariantCulture);
+            }
+            
             subNode = node.SelectSingleNode("BDOpenIn");
             if (subNode != null)
             {
@@ -9198,6 +9206,7 @@ $HorzAlign          =   Center
                 textWriter.WriteElementString("OcrTrainFonts", settings.Tools.OcrTrainFonts);
                 textWriter.WriteElementString("OcrTrainMergedLetters", settings.Tools.OcrTrainMergedLetters);
                 textWriter.WriteElementString("OcrTrainSrtFile", settings.Tools.OcrTrainSrtFile);
+                textWriter.WriteElementString("OcrUseWordSplitList", settings.Tools.OcrUseWordSplitList.ToString(CultureInfo.InvariantCulture));
                 textWriter.WriteElementString("BDOpenIn", settings.Tools.BDOpenIn);
                 textWriter.WriteElementString("Interjections", settings.Tools.Interjections);
                 textWriter.WriteElementString("MicrosoftBingApiId", settings.Tools.MicrosoftBingApiId);

src/libse/Dictionaries/StringWithoutSpaceSplitToWords.cs

@@ -1,6 +1,7 @@
 using System;
 using System.Collections.Generic;
 using System.Linq;
+using Nikse.SubtitleEdit.Core.Common;
 
 namespace Nikse.SubtitleEdit.Core.Dictionaries
 {
@@ -8,6 +9,11 @@ namespace Nikse.SubtitleEdit.Core.Dictionaries
     {
         public static string SplitWord(string[] words, string input)
         {
+            if (!Configuration.Settings.Tools.OcrUseWordSplitList)
+            {
+                return input;
+            }
+
             var usedWords = new List<string>();
             var result = SplitWord(words, input, string.Empty, usedWords);
             if (result != input)