From 86fb003192f5b156e819d8295c91a4d47cab16db Mon Sep 17 00:00:00 2001
From: niksedk <nikse.dk@gmail.com>
Date: Fri, 21 Jan 2022 19:44:40 +0100
Subject: [PATCH] Update dictionaries

---
 Dictionaries/en_US_user.xml                      | 10 ++++++++++
 Dictionaries/en_names.xml                        |  2 ++
 Dictionaries/names.xml                           |  2 ++
 src/ui/Forms/Ocr/WordSplitDictionaryGenerator.cs | 10 ++++++++--
 4 files changed, 22 insertions(+), 2 deletions(-)

diff --git a/Dictionaries/en_US_user.xml b/Dictionaries/en_US_user.xml
index 65d4955f6..2c8d99a50 100644
--- a/Dictionaries/en_US_user.xml
+++ b/Dictionaries/en_US_user.xml
@@ -198,11 +198,13 @@
   <word>financials</word>
   <word>flavour</word>
   <word>flavours</word>
+  <word>floricultural</word>
   <word>fluorine</word>
   <word>flushin'</word>
   <word>flyer</word>
   <word>forevermore</word>
   <word>francium</word>
+  <word>frenemy</word>
   <word>fuckable</word>
   <word>fundraiser</word>
   <word>gadolinium</word>
@@ -229,6 +231,7 @@
   <word>hashtag</word>
   <word>hassium</word>
   <word>hatchling</word>
+  <word>hawala</word>
   <word>heh</word>
   <word>helium</word>
   <word>heloooooo</word>
@@ -356,6 +359,8 @@
   <word>parents'</word>
   <word>percutaneously</word>
   <word>perfusionist</word>
+  <word>pescatarian</word>
+  <word>pescatarians</word>
   <word>phosphorus</word>
   <word>photoshopped</word>
   <word>piccaninny</word>
@@ -436,6 +441,7 @@
   <word>splenectomy</word>
   <word>spork</word>
   <word>spotteth</word>
+  <word>squirrelly</word>
   <word>stenosis</word>
   <word>stent</word>
   <word>stenting</word>
@@ -450,6 +456,8 @@
   <word>sulfur</word>
   <word>sulphur</word>
   <word>sunglass</word>
+  <word>supervillain</word>
+  <word>supervillains</word>
   <word>supervolcano</word>
   <word>synchronicity</word>
   <word>syncopal</word>
@@ -512,8 +520,10 @@
   <word>voicemail</word>
   <word>voila</word>
   <word>walkthrough</word>
+  <word>weaponized</word>
   <word>weirding</word>
   <word>what'd</word>
+  <word>what'll</word>
   <word>what're</word>
   <word>when'd</word>
   <word>where'd</word>
diff --git a/Dictionaries/en_names.xml b/Dictionaries/en_names.xml
index c942d0f06..ac4e17e0e 100644
--- a/Dictionaries/en_names.xml
+++ b/Dictionaries/en_names.xml
@@ -1531,6 +1531,7 @@
   <name>Peruvians</name>
   <name>Philippines</name>
   <name>Pinella</name>
+  <name>Pippi Longstocking</name>
   <name>Pinkman</name>
   <name>Pitcairn Islands</name>
   <name>Poindexter</name>
@@ -1585,6 +1586,7 @@
   <name>Rayne</name>
   <name>Rebekah</name>
   <name>Reece</name>
+  <name>Reddington</name>
   <name>Regan</name>
   <name>Reid</name>
   <name>Remington</name>
diff --git a/Dictionaries/names.xml b/Dictionaries/names.xml
index a5ec3a446..2c7cca656 100644
--- a/Dictionaries/names.xml
+++ b/Dictionaries/names.xml
@@ -2561,6 +2561,7 @@ This file is case sensitive.
   <name>Idris Elba</name>
   <name>Igor</name>
   <name>III</name>
+  <name>Ikaris</name>
   <name>Ikea</name>
   <name>Ilana</name>
   <name>Ilithyia</name>
@@ -3918,6 +3919,7 @@ This file is case sensitive.
   <name>Munich</name>
   <name>Muray</name>
   <name>Murdoch</name>
+  <name>Murdock</name>
   <name>Muriel</name>
   <name>Murmillo</name>
   <name>Murphy</name>
diff --git a/src/ui/Forms/Ocr/WordSplitDictionaryGenerator.cs b/src/ui/Forms/Ocr/WordSplitDictionaryGenerator.cs
index bf3fc9878..a2626d423 100644
--- a/src/ui/Forms/Ocr/WordSplitDictionaryGenerator.cs
+++ b/src/ui/Forms/Ocr/WordSplitDictionaryGenerator.cs
@@ -14,7 +14,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
 {
     public partial class WordSplitDictionaryGenerator : Form
     {
-        private List<Subtitle> _subtitleList;
+        private readonly List<Subtitle> _subtitleList;
         private Hunspell _hunspell;
 
         public WordSplitDictionaryGenerator()
@@ -103,7 +103,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                 if (!FileUtil.IsBluRaySup(fileName) && !FileUtil.IsVobSub(fileName) &&
                     !((ext == ".mkv" || ext == ".mks") && FileUtil.IsMatroskaFile(fileName)))
                 {
-                    SubtitleFormat format = sub.LoadSubtitle(fileName, out _, null);
+                    var format = sub.LoadSubtitle(fileName, out _, null);
 
                     if (format == null)
                     {
@@ -151,6 +151,12 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
             {
                 foreach (var p in subtitle.Paragraphs)
                 {
+                    if (p.Text.Contains("Synced and corrected by", StringComparison.OrdinalIgnoreCase) ||
+                        p.Text.Contains("www."))
+                    {
+                        continue;
+                    }
+
                     var words = SpellCheckWordLists.Split(HtmlUtil.RemoveHtmlTags(p.Text, true));
                     foreach (var word in words)
                     {