Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-22 03:02:35 +01:00
Code Issues Projects Releases Wiki Activity

Minor fixes for OCR

Browse Source 
Handle "1" as "I" i some situations + don't count "I" and "a" as wrong letters in English
This commit is contained in:
Nikolaj Olsson 2020-04-24 12:05:53 +02:00
parent 11820b2273
commit a3e42a4026
2 changed files with 9 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Dictionaries/eng_OCRFixReplaceList.xml
View File

@ -173,6 +173,7 @@
<Word from="AIso" to="Also" />
<Word from="AIVIBULANCE" to="AMBULANCE" />
<Word from="AIways" to="Always" />
<Word from="aimost" to="almost" />
<Word from="ajanitor" to="a janitor" />
<Word from="ajealous" to="a jealous" />
<Word from="ajob" to="a job" />
@ -2201,6 +2202,7 @@
<Word from="theysay" to="they say" />
<Word from="theyve" to="they've" />
<Word from="Theyve" to="They've" />
<Word from="thing'II" to="thing'll" />
<Word from="Thinkbackto" to="Think back to" />
<Word from="thinkl" to="think l" />
<Word from="thinkofafew" to="think of a few" />
@ -2380,6 +2382,7 @@
<Word from="we,re" to="we're" />
<Word from="we,ve" to="we've" />
<Word from="We,ve" to="We've" />
<Word from="We'il" to="We'll" />
<Word from="We//" to="Well" />
<Word from="We/came" to="Welcome" />
<Word from="we/come" to="welcome" />
@ -3201,5 +3204,6 @@
<RegEx find="^_\.\.(\p{L})" replaceWith="...$1" />
<RegEx find=" l([!?\.])" replaceWith=" I$1" />
<RegEx find="\b\|\b" replaceWith="I" />
<RegEx find="\b1 (know|will|almost|didn't|get|got|have|apologize|paid|like|think|would|hope|shall|chose|choose|won|am|was|don't|just|start|run|saw|said|believe|try|ever|need|certainly|can't|anticipated|did|can|rang|heard|gave|came|decided|should|took|wanted|read|thought|was|still|do|love|want|overstepped|accept|authorized|owe|understand|made|guess|bumped|wasn't|mean|admire|had|spent|told|see|walk|were|help|definitely|could|say|take|brought|assume|proposed|realized|loved|base|left|change|changed|rule|feel|date|dated|imagine|went|kind|couldn't|wouldn't|work|care|make|lost)+\b" replaceWith="I $1" />
</RegularExpressions>
</OCRFixReplaceList>

5
src/Logic/Ocr/OcrFixEngine.cs
View File

@ -2015,6 +2015,11 @@ namespace Nikse.SubtitleEdit.Logic.Ocr
correct = word.Length > 2 && _hunspell.Spell(word.Trim('\''));
}
if (!correct && word.Length == 1 && _threeLetterIsoLanguageName == "eng" && (word == "I" || word == "A" || word == "a"))
{
correct = true;
}
if (correct)
{
numberOfCorrectWords++;