Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-10-27 14:32:35 +01:00
Code Issues Projects Releases Wiki Activity

Merge pull request #7673 from diomed/master

Browse Source 
Croatian OCR fixes
This commit is contained in:
Nikolaj Olsson 2023-11-28 07:41:28 +01:00 committed by GitHub
commit 6b15cc76d4
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 30 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
Dictionaries/hrv_OCRFixReplaceList.xml
View File

@ -128,6 +128,7 @@
<Word from="budep" to="budeš" />
<Word from="budilnkik" to="budilica" />
<Word from="buđav" to="pljesniv" />
<Word from="buldožder" to="buldožer" />
<!-- c -->
<Word from="cedila" to="cijedila" />
<Word from="cedilu" to="cjedilu" />
@ -2381,6 +2382,7 @@
<Word from="Talasi" to="Valovi" />
<Word from="talasima" to="valovima" />
<Word from="Talasima" to="Valovima" />
<Word from="taoc" to="talac" />
<Word from="teba" to="treba" />
<Word from="tebalo" to="trebalo" />
<Word from="tečna" to="tekuća" />
@ -2903,6 +2905,7 @@
<Word from="Brižit" to="Brigitte" />
<Word from="Bruks" to="Brooks" />
<Word from="Buč" to="Butch" />
<Word from="Cuba" to="Kuba" />
<Word from="Siti" to="City" />
<Word from="Sitija" to="Cityja" />
<Word from="Čarli" to="Charlie" />
@ -3640,10 +3643,28 @@
<Word from="živjell" to="živjeli" />
<Word from="žlvih" to="živih" />
</WholeWords>
<PartialWordsAlways />
<PartialWords />
<PartialWordsAlways>
<WordPart from="¤" to="o" />
<WordPart from="IVI" to="M" />
<WordPart from="lVI" to="M" />
<WordPart from="IVl" to="M" />
<WordPart from="lVl" to="M" />
</PartialWordsAlways>
<PartialWords>
<WordPart from="fi" to="fi" />
<WordPart from="fl" to="fl" />
<WordPart from="/" to="l" />
<WordPart from="vv" to="w" />
</PartialWords>
<WholeLines />
<PartialLinesAlways />
<PartialLinesAlways>
<LinePart from=",.," to="..." />
<LinePart from=" l " to=" I " />
<LinePart from=" l." to=" I." />
<LinePart from=" l?" to=" I?" />
<LinePart from=" l!" to=" I!" />
<LinePart from=" . " to=". " />
</PartialLinesAlways>
<PartialLines>
<LinePart from="Ako ej" to="Ako je" />
<LinePart from="ako ej" to="ako je" />
@ -3831,6 +3852,7 @@
<LinePart from="došli smo ovde" to="došli smo ovamo" />
<LinePart from="Što ej" to="Što je" />
<LinePart from="što ej" to="što je" />
<LinePart from="što šta" to="štošta" />
<LinePart from="to da ponavljaš" to="to ponavljati" />
<LinePart from="to j" to="to je" />
<LinePart from="to ej" to="to je" />
@ -3953,6 +3975,7 @@
<RegEx find="boži[čć]([aeiu]|em|ima)?\b" replaceWith="Božić$1" />
<RegEx find="(?&lt;!\A|[.!?][&quot;”’]?\s+)\bBoži[čć]n([aeiou]|om|im)\b" replaceWith="božićn$1" />
<RegEx find="branb" replaceWith="bramb" />
<RegEx find="bsor" replaceWith="psor" />
<RegEx find="[bB]ubašvab" replaceWith="žohar" />
<RegEx find="bukval" replaceWith="doslov" />
<RegEx find="Bukval" replaceWith="Doslov" />
@ -4837,6 +4860,8 @@
<RegEx find="bqr" replaceWith="bar" />
<RegEx find="bs([to])" replaceWith="ps$1" />
<RegEx find="bunić" replaceWith="bunit ć" />
<!-- opservacija-->
<RegEx find="bzer" replaceWith="pser" />
<RegEx find="iliše" replaceWith="ilizira" />
<RegEx find="ilišu" replaceWith="iliziraju" />
<RegEx find="izc" replaceWith="isc" />
@ -5139,6 +5164,7 @@
<RegEx find="seden" replaceWith="sjeden" />
<RegEx find="selektuj" replaceWith="odaber" />
<RegEx find="shvt" replaceWith="shvat" />
<RegEx find="slk" replaceWith="slj" />
<RegEx find="sljen" replaceWith="slen" />
<RegEx find="sjp" replaceWith="spj" />
<RegEx find="smeva" replaceWith="smijava" />
@ -5201,6 +5227,7 @@
<RegEx find="trišu" replaceWith="triraju" />
<RegEx find="trezni" replaceWith="trijezni" />
<RegEx find="uas" replaceWith="zas" />
<RegEx find="udep" replaceWith="udeš" />
<RegEx find="uej" replaceWith="uje" />
<RegEx find="ueln" replaceWith="ualn" />
<RegEx find="uisa" replaceWith="uira" />