From 1026af5cfc87cd121e016f8d59dc25ba7feb1177 Mon Sep 17 00:00:00 2001
From: niksedk <nikse.dk@gmail.com>
Date: Wed, 3 Aug 2016 13:31:33 +0200
Subject: [PATCH] Minor fix for OCR unknown word (line split)

---
 src/Forms/OCRSpellCheck.cs    | 2 +-
 src/Logic/OCR/OcrFixEngine.cs | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/Forms/OCRSpellCheck.cs b/src/Forms/OCRSpellCheck.cs
index 3df63d48a..119a7df0d 100644
--- a/src/Forms/OCRSpellCheck.cs
+++ b/src/Forms/OCRSpellCheck.cs
@@ -88,7 +88,7 @@ namespace Nikse.SubtitleEdit.Forms
         {
             if (word != null && richTextBoxParagraph.Text.Contains(word))
             {
-                const string ExpectedWordBoundaryChars = " <>-\"”“[]'‘`´¶()♪¿¡.…—!?,:;/\r\n";
+                const string ExpectedWordBoundaryChars = " <>-\"”“«»[]'‘`´¶()♪¿¡.…—!?,:;/\r\n";
                 for (int i = 0; i < richTextBoxParagraph.Text.Length; i++)
                 {
                     if (richTextBoxParagraph.Text.Substring(i).StartsWith(word))
diff --git a/src/Logic/OCR/OcrFixEngine.cs b/src/Logic/OCR/OcrFixEngine.cs
index 0292b1462..0c40307a8 100644
--- a/src/Logic/OCR/OcrFixEngine.cs
+++ b/src/Logic/OCR/OcrFixEngine.cs
@@ -959,7 +959,7 @@ namespace Nikse.SubtitleEdit.Logic.Ocr
                 }
             }
 
-            string[] words = tempLine.Replace("</i>", string.Empty).Split((Environment.NewLine + " ¡¿,.!?:;()[]{}+-£\"”“#&%…—♪").ToCharArray(), StringSplitOptions.RemoveEmptyEntries);
+            string[] words = tempLine.Replace("</i>", string.Empty).Split((Environment.NewLine + " ¡¿,.!?:;()[]{}+-£\"”“«»#&%…—♪").ToCharArray(), StringSplitOptions.RemoveEmptyEntries);
             for (int i = 0; i < words.Length; i++)
             {
                 string word = words[i].TrimStart('\'');
@@ -1382,7 +1382,7 @@ namespace Nikse.SubtitleEdit.Logic.Ocr
                 return 0;
 
             int wordsNotFound = 0;
-            var words = HtmlUtil.RemoveOpenCloseTags(line, HtmlUtil.TagItalic).Split((Environment.NewLine + " ¡¿,.!?:;()[]{}+-$£\"#&%…“”").ToCharArray(), StringSplitOptions.RemoveEmptyEntries);
+            var words = HtmlUtil.RemoveOpenCloseTags(line, HtmlUtil.TagItalic).Split((Environment.NewLine + " ¡¿,.!?:;()[]{}+-$£\"#&%…“”«»").ToCharArray(), StringSplitOptions.RemoveEmptyEntries);
             for (int i = 0; i < words.Length; i++)
             {
                 string word = words[i];