Merge branch 'master' of https://github.com/MrFlapstaart/GameOCRTTS

OCR.cs

@@ -32,13 +32,9 @@ internal static TextBlock GetTextFromTiffStream(byte[] image)
 
                         ComposedBlock cblock = result.PrintSpace.ComposedBlock.OrderByDescending(x => x.WordsInComposedBlock).FirstOrDefault() ?? new ComposedBlock();
                         TextBlock block = cblock.Blocks.OrderByDescending(x => x.WordsInBlock).FirstOrDefault() ?? new TextBlock();
-
-                        Regex rgx = new Regex("[^a-zA-Z0-9.,'?!: -]");
-                        foreach (var line in block.Lines)
-                        {
-                            line.Words.RemoveAll(x => string.IsNullOrEmpty(rgx.Replace(x.Content.Replace("|", "I"), "")));
-                        }
-                        block.Lines.RemoveAll(x => x.WordsInLine == 0);
+                        block = TextHelper.ProcessTextBlock(block);
+                        if (block.Lines.Count > 0)
+                            ResetBlockBoundaries(block);
 
                         return block;                            
 
@@ -51,5 +47,21 @@ internal static TextBlock GetTextFromTiffStream(byte[] image)
                 return new TextBlock() { Text = $"Error: {ex.Message}" };
             }            
         }
+
+        private static void ResetBlockBoundaries(TextBlock block)
+        {
+            foreach (var line in block.Lines)
+            {
+                line.HPos = line.Words.Min(x => x.HPos);
+                line.VPos = line.Words.Min(x => x.VPos);
+                line.Width = line.Words.Max(x => x.HPos + x.Width) - line.HPos;
+                line.Height = line.Words.Max(x => x.VPos + x.Height) - line.VPos;
+            }
+
+            block.HPos = block.Lines.Min(x => x.HPos);
+            block.VPos = block.Lines.Min(x => x.VPos);
+            block.Width = block.Lines.Max(x => x.HPos + x.Width) - block.HPos + 1;
+            block.Height = block.Lines.Max(x => x.VPos + x.Height) - block.VPos + 1;
+        }
     }
 }

OCRResult.cs

@@ -65,7 +65,7 @@ public string Text
 
         public override string ToString()
         {
-			return $"{VPos},{Height} {Text}";
+			return $"{HPos},{VPos},{Width},{Height} : {Text}";
         }
 
     }

TextHelper.cs

@@ -1,4 +1,5 @@
-using System.Text;
+using System;
+using System.Text;
 using System.Text.RegularExpressions;
 
 namespace GameOCRTTS
@@ -54,6 +55,50 @@ internal static string RemoveGarbageText(string text)
             return result.ToString();
         }
 
+        internal static TextBlock ProcessTextBlock(TextBlock block)
+        {
+            Regex rgxspec = new Regex("[^a-zA-Z0-9.,'?!: -]");
+            foreach (var line in block.Lines)
+            {
+                line.Words.RemoveAll(x => string.IsNullOrEmpty(rgxspec.Replace(x.Content.Replace("|", "I"), "").Trim()));
+                line.Words.RemoveAll(x => !IsValidWord(x.Content));                
+            }
+            block.Lines.RemoveAll(x => x.WordsInLine == 1 && x.Text.Length <= 2);
+            block.Lines.RemoveAll(x => x.WordsInLine == 0);
+
+            return block;
+        }
+
+        private static bool IsValidWord(string word)
+        {
+            Regex rgx = new Regex("[^a-zA-Z0-9']");
+            Regex rgxnum = new Regex("[^0-9]");
+            Regex rgxvowel = new Regex("[^aeiouy]");
+
+            string strip = rgx.Replace(word, "")?.ToLower()?.Trim();
+            if (string.IsNullOrEmpty(strip))
+                return false;
+
+            if (strip.Length == 1 && strip != "i" && strip != "a")
+                return false;
+
+            string numbers = rgxnum.Replace(word, "");
+            bool onlynumbers = numbers.Length == strip.Length && numbers.Length > 0;            
+            int vowels = rgxvowel.Replace(strip, "").Length;
+
+            if (strip.Length == 1 && !onlynumbers && strip != "i" && strip != "a")
+                return false;
+
+
+            if (!word.StartsWith("..") && strip.Length <= 5 && strip.Length > 1 && !onlynumbers && !strip.StartsWith("hm"))
+            {
+                if (vowels == 0 || vowels == strip.Length)
+                    return false;
+            }
+
+            return true;
+        }
+
         internal static string StripSpecialCharacters(string text)
         {
             if (string.IsNullOrEmpty(text))