Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-23 11:42:36 +01:00
Code Issues Projects Releases Wiki Activity

Merge pull request #3236 from diomed/patch-1

Browse Source 
Update Croatian/Bosnian OCR
This commit is contained in:
Nikolaj Olsson 2018-12-13 15:47:37 +01:00 committed by GitHub
commit 74745a650a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 16 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Dictionaries/ba_OCRFixReplaceList.xml
View File

@ -1229,6 +1229,7 @@
<RegEx find="c([ae])sc" replaceWith="č$1šć" />
<RegEx find="cek" replaceWith="ček" />
<RegEx find="Ček" replaceWith="Ček" />
<RegEx find="ce[sš]c" replaceWith="češć" />
<RegEx find="[cC]etvrt" replaceWith="četvrt" />
<RegEx find="cetv" replaceWith="četv" />
<RegEx find="Cetv" replaceWith="Četv" />
@ -1723,6 +1724,7 @@
<RegEx find="gusi" replaceWith="guši" />
<RegEx find="(?&lt;!\b[oO]zlo|\b[iI]sp(rip)?ov|i)jedjen" replaceWith="ijeđen" />
<RegEx find="hic(k|en)" replaceWith="hić$1" />
<RegEx find="hodj" replaceWith="hođ" />
<RegEx find="hva[cč]a" replaceWith="hvaća" />
<RegEx find="hva[cč]as" replaceWith="hvaćaš" />
<RegEx find="hva([lt])is" replaceWith="hva$1iš" />

14
Dictionaries/hrv_OCRFixReplaceList.xml
View File

@ -71,14 +71,19 @@
<Word from="bitci" to="bitki" />
<Word from="bled" to="blijed" />
<Word from="blede" to="blijede" />
<Word from="boksujem" to="boksam" />
<Word from="Boksujem" to="Boksam" />
<Word from="boksuješ" to="boksaš" />
<Word from="Boksuješ" to="Boksaš" />
<Word from="bolesan" to="bolestan" />
<Word from="bolila" to="boljela" />
<Word from="borićeš" to="borit ćeš" />
<Word from="Borićeš" to="Borit ćeš" />
<Word from="Boriću" to="Borit ću" />
<Word from="braon" to="smeđa" />
<Word from="bregu" to="brijegu" />
<Word from="bti" to="biti" />
<Word from="cedila" to="cijedila" />
<Word from="cedilu" to="cjedilu" />
<Word from="ceo" to="cijeli" />
<Word from="Ceo" to="Cijeli" />
@ -277,6 +282,7 @@
<Word from="familijama" to="obiteljima" />
<Word from="familiji" to="obitelji" />
<Word from="farmerke" to="traperice" />
<Word from="Farmerke" to="Traperice" />
<Word from="farmerkama" to="trapericama" />
<Word from="flertuje" to="očijuka" />
<Word from="foka" to="tuljan" />
@ -526,6 +532,7 @@
<Word from="koverti" to="kuverti" />
<Word from="kovertu" to="kuvertu" />
<Word from="kritikuju" to="kritiziraju" />
<Word from="Krst" to="Križ" />
<Word from="krst" to="križ" />
<Word from="krsta" to="križa" />
<Word from="krstaši" to="križari" />
@ -721,6 +728,7 @@
<Word from="Negde" to="Negdje" />
<Word from="neguje" to="njeguje" />
<Word from="neguju" to="njeguju" />
<Word from="neito" to="nešto" />
<Word from="nekto" to="netko" />
<Word from="nemi" to="nijemi" />
<Word from="nemrem" to="ne mogu" />
@ -1164,6 +1172,7 @@
<Word from="Rjekom" to="Rijekom" />
<Word from="Rjekama" to="Rijekama" />
<Word from="rengen" to="rendgen" />
<Word from="reoe" to="reče" />
<Word from="repuješ" to="repaš" />
<Word from="resetuje" to="resetira" />
<Word from="resetujte" to="resetirajte" />
@ -1496,6 +1505,7 @@
<Word from="Udata" to="Udana" />
<Word from="udatoj" to="udanoj" />
<Word from="udeo" to="udio" />
<Word from="udelim" to="udijelim" />
<Word from="ugljenik" to="ugljik" />
<Word from="uglova" to="kutova" />
<Word from="uliva" to="ulijeva" />
@ -2152,6 +2162,8 @@
<LinePart from="možeš da budeš" to="možeš biti" />
<LinePart from="može da diše" to="može disati" />
<LinePart from="možeš da dobiješ" to="možeš dobiti" />
<LinePart from="može da uđe" to="može ući" />
<LinePart from="možeš da uđeš" to="možeš ući" />
<LinePart from="možemo da imamo" to="možemo imati" />
<LinePart from="na večer" to="navečer" />
<LinePart from="Na večer" to="Navečer" />
@ -2215,6 +2227,7 @@
<LinePart from="želim da idem" to="želim ići" />
<LinePart from="želim da odem" to="želim otići" />
<LinePart from="želiš da odeš" to="želiš otići" />
<LinePart from="Želiš da učiniš" to="Želiš učiniti" />
<LinePart from="žele da uđu" to="žele ući" />
<LinePart from="želiš da uđeš" to="želiš ući" />
<LinePart from="želim da umrem" to="želim umrijeti" />
@ -2684,6 +2697,7 @@
<RegEx find="(?&lt;![mn])o(d?)se([kć])(?!s)" replaceWith="o$1sje$2" />
<RegEx find="odu?hvat" replaceWith="othvat" />
<RegEx find="(?&lt;!r)oent" replaceWith="oant" />
<RegEx find="ovren" replaceWith="ovoren" />
<RegEx find="([pP])ogrj?eši(?!v)" replaceWith="$1ogriješi" />
<RegEx find="([pP])olen" replaceWith="$1elud" />
<RegEx find="lude([lt])" replaceWith="ludje$1" />