Update dictionaries

This commit is contained in:
niksedk 2022-01-21 19:44:40 +01:00
parent ccda5f2084
commit 86fb003192
4 changed files with 22 additions and 2 deletions

View File

@ -198,11 +198,13 @@
<word>financials</word> <word>financials</word>
<word>flavour</word> <word>flavour</word>
<word>flavours</word> <word>flavours</word>
<word>floricultural</word>
<word>fluorine</word> <word>fluorine</word>
<word>flushin'</word> <word>flushin'</word>
<word>flyer</word> <word>flyer</word>
<word>forevermore</word> <word>forevermore</word>
<word>francium</word> <word>francium</word>
<word>frenemy</word>
<word>fuckable</word> <word>fuckable</word>
<word>fundraiser</word> <word>fundraiser</word>
<word>gadolinium</word> <word>gadolinium</word>
@ -229,6 +231,7 @@
<word>hashtag</word> <word>hashtag</word>
<word>hassium</word> <word>hassium</word>
<word>hatchling</word> <word>hatchling</word>
<word>hawala</word>
<word>heh</word> <word>heh</word>
<word>helium</word> <word>helium</word>
<word>heloooooo</word> <word>heloooooo</word>
@ -356,6 +359,8 @@
<word>parents'</word> <word>parents'</word>
<word>percutaneously</word> <word>percutaneously</word>
<word>perfusionist</word> <word>perfusionist</word>
<word>pescatarian</word>
<word>pescatarians</word>
<word>phosphorus</word> <word>phosphorus</word>
<word>photoshopped</word> <word>photoshopped</word>
<word>piccaninny</word> <word>piccaninny</word>
@ -436,6 +441,7 @@
<word>splenectomy</word> <word>splenectomy</word>
<word>spork</word> <word>spork</word>
<word>spotteth</word> <word>spotteth</word>
<word>squirrelly</word>
<word>stenosis</word> <word>stenosis</word>
<word>stent</word> <word>stent</word>
<word>stenting</word> <word>stenting</word>
@ -450,6 +456,8 @@
<word>sulfur</word> <word>sulfur</word>
<word>sulphur</word> <word>sulphur</word>
<word>sunglass</word> <word>sunglass</word>
<word>supervillain</word>
<word>supervillains</word>
<word>supervolcano</word> <word>supervolcano</word>
<word>synchronicity</word> <word>synchronicity</word>
<word>syncopal</word> <word>syncopal</word>
@ -512,8 +520,10 @@
<word>voicemail</word> <word>voicemail</word>
<word>voila</word> <word>voila</word>
<word>walkthrough</word> <word>walkthrough</word>
<word>weaponized</word>
<word>weirding</word> <word>weirding</word>
<word>what'd</word> <word>what'd</word>
<word>what'll</word>
<word>what're</word> <word>what're</word>
<word>when'd</word> <word>when'd</word>
<word>where'd</word> <word>where'd</word>

View File

@ -1531,6 +1531,7 @@
<name>Peruvians</name> <name>Peruvians</name>
<name>Philippines</name> <name>Philippines</name>
<name>Pinella</name> <name>Pinella</name>
<name>Pippi Longstocking</name>
<name>Pinkman</name> <name>Pinkman</name>
<name>Pitcairn Islands</name> <name>Pitcairn Islands</name>
<name>Poindexter</name> <name>Poindexter</name>
@ -1585,6 +1586,7 @@
<name>Rayne</name> <name>Rayne</name>
<name>Rebekah</name> <name>Rebekah</name>
<name>Reece</name> <name>Reece</name>
<name>Reddington</name>
<name>Regan</name> <name>Regan</name>
<name>Reid</name> <name>Reid</name>
<name>Remington</name> <name>Remington</name>

View File

@ -2561,6 +2561,7 @@ This file is case sensitive.
<name>Idris Elba</name> <name>Idris Elba</name>
<name>Igor</name> <name>Igor</name>
<name>III</name> <name>III</name>
<name>Ikaris</name>
<name>Ikea</name> <name>Ikea</name>
<name>Ilana</name> <name>Ilana</name>
<name>Ilithyia</name> <name>Ilithyia</name>
@ -3918,6 +3919,7 @@ This file is case sensitive.
<name>Munich</name> <name>Munich</name>
<name>Muray</name> <name>Muray</name>
<name>Murdoch</name> <name>Murdoch</name>
<name>Murdock</name>
<name>Muriel</name> <name>Muriel</name>
<name>Murmillo</name> <name>Murmillo</name>
<name>Murphy</name> <name>Murphy</name>

View File

@ -14,7 +14,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
{ {
public partial class WordSplitDictionaryGenerator : Form public partial class WordSplitDictionaryGenerator : Form
{ {
private List<Subtitle> _subtitleList; private readonly List<Subtitle> _subtitleList;
private Hunspell _hunspell; private Hunspell _hunspell;
public WordSplitDictionaryGenerator() public WordSplitDictionaryGenerator()
@ -103,7 +103,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
if (!FileUtil.IsBluRaySup(fileName) && !FileUtil.IsVobSub(fileName) && if (!FileUtil.IsBluRaySup(fileName) && !FileUtil.IsVobSub(fileName) &&
!((ext == ".mkv" || ext == ".mks") && FileUtil.IsMatroskaFile(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) if (format == null)
{ {
@ -151,6 +151,12 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
{ {
foreach (var p in subtitle.Paragraphs) 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)); var words = SpellCheckWordLists.Split(HtmlUtil.RemoveHtmlTags(p.Text, true));
foreach (var word in words) foreach (var word in words)
{ {