Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-23 03:33:18 +01:00
Code Issues Projects Releases Wiki Activity

Merge pull request #2970 from diomed/patch-1

Browse Source 
Cro & Ba OCR Fixes
This commit is contained in:
Nikolaj Olsson 2018-07-04 09:12:43 +02:00 committed by GitHub
commit c5077ed6e9
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 35 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
Dictionaries/ba_OCRFixReplaceList.xml
View File

@ -1209,8 +1209,8 @@
<RegEx find="cas([au]|om|ovima)\b" replaceWith="sat$1" />
<RegEx find="Cas([au]|om|ovima)\b" replaceWith="Sat$1" />
<RegEx find="[cC]ascen" replaceWith="čašćen" />
<RegEx find="[cC]ek" replaceWith="ček" />
<RegEx find="[cC]ekov" replaceWith="čekov" />
<RegEx find="cek" replaceWith="ček" />
<RegEx find="Ček" replaceWith="Ček" />
<RegEx find="[cC]etvrt" replaceWith="četvrt" />
<RegEx find="cetv" replaceWith="četv" />
<RegEx find="Cetv" replaceWith="Četv" />
@ -1228,7 +1228,7 @@
<RegEx find="[cC]udn" replaceWith="čudn" />
<RegEx find="[cCČč]udovi[sš]" replaceWith="čudoviš" />
<RegEx find="\b[cC]uje" replaceWith="čuje" />
<RegEx find="(?&lt;!a)cuje" replaceWith="čuje" />
<RegEx find="(?&lt;![ba])cuje" replaceWith="čuje" />
<RegEx find="[cC]u([vl])a" replaceWith="ču$1a" />
<RegEx find="cut([ei])" replaceWith="šut$1" />
<RegEx find="Cut([ei])" replaceWith="Šut$1" />
@ -1750,8 +1750,8 @@
<RegEx find="nisten" replaceWith="ništen" />
<RegEx find="(?&lt;![Ee]t|[Gg]e|[Rr]i|[Tt]am)noc" replaceWith="noć" />
<RegEx find="oceju" replaceWith="oće" />
<RegEx find="oceo" replaceWith="očeo" />
<RegEx find="ocni" replaceWith="očni" />
<RegEx find="([Oo])ce([ko])" replaceWith="$1če$2" />
<RegEx find="oc[mn]i" replaceWith="očni" />
<RegEx find="o([mk])aze" replaceWith="o$1aže" />
<RegEx find="/\bni[cć]k" replaceWith="ničk" />
<RegEx find="nosc" replaceWith="nošć" />
@ -1776,6 +1776,7 @@
<RegEx find="ronad" replaceWith="ronađ" />
<RegEx find="([KkPp])renes\b" replaceWith="$1reneš" />
<RegEx find="sapni" replaceWith="šapni" />
<RegEx find="sce([zk])" replaceWith="šče$1" />
<RegEx find="setnj" replaceWith="šetnj" />
<RegEx find="slusa" replaceWith="sluša" />
<RegEx find="slusk" replaceWith="slušk" />

29
Dictionaries/hrv_OCRFixReplaceList.xml
View File

@ -36,6 +36,7 @@
<Word from="begu" to="bijegu" />
<Word from="begstva" to="bijega" />
<Word from="bejaše" to="bijaše" />
<Word from="bekstva" to="bijega" />
<Word from="bekstvo" to="bijeg" />
<Word from="bekstvu" to="bijegu" />
<Word from="begstvu" to="bijegu" />
@ -190,6 +191,7 @@
<Word from="dospeju" to="dospiju" />
<Word from="dođavola" to="dovraga" />
<Word from="Dođavola" to="Dovraga" />
<Word from="drajv" to="drive" />
<Word from="drug" to="prijatelj" />
<Word from="drugari" to="prijatelji" />
<Word from="drugare" to="prijatelje" />
@ -657,6 +659,7 @@
<Word from="pantalonama" to="hlačama" />
<Word from="parčadi" to="komadića" />
<Word from="parče" to="komadić" />
<Word from="Parče" to="Komadić" />
<Word from="parčeta" to="komadića" />
<Word from="parčetu" to="komadiću" />
<Word from="parčetom" to="komadićem" />
@ -954,10 +957,12 @@
<Word from="smeškom" to="smješkom" />
<Word from="smjeo" to="smio" />
<Word from="smrdeo" to="smrdio" />
<Word from="smrdelo" to="smrdjelo" />
<Word from="sintersajzer" to="synthesizer" />
<Word from="sitnisajzer" to="synthesizer" />
<Word from="skelet" to="kostur" />
<Word from="sočiva" to="leće" />
<Word from="Sočiva" to="Leće" />
<Word from="sočivima" to="lećama" />
<Word from="sočivo" to="leća" />
<Word from="sočivom" to="lećom" />
@ -970,11 +975,14 @@
<Word from="spasen" to="spašen" />
<Word from="spasla" to="spasila" />
<Word from="spasli" to="spasili" />
<Word from="spavaćeš" to="spavat ćeš" />
<Word from="speluje" to="sriče" />
<Word from="speluju" to="sriču" />
<Word from="spojiće" to="spojit će" />
<Word from="spolja" to="izvana" />
<Word from="Srediće" to="Sredit će" />
<Word from="Sredićemo" to="Sredit ćemo" />
<Word from="sredićemo" to="sredit ćemo" />
<Word from="Srediću" to="Sredit ću" />
<!-- nisu za regex - ne odgovaraju zbog dodatnog ju na kraju -->
<Word from="stabilizuju" to="stabiliziraju" />
@ -1080,6 +1088,7 @@
<Word from="toha" to="toga" />
<Word from="tovj" to="tvoj" />
<Word from="trabam" to="trebam" />
<Word from="trezna" to="trijezna" />
<Word from="trkanje" to="utrkivanje" />
<Word from="trkanja" to="utrkivanja" />
<Word from="trkao" to="utrkivao" />
@ -1249,6 +1258,8 @@
<Word from="zaceljivanje" to="zacjeljivanje" />
<Word from="zagreva" to="zagrijava" />
<Word from="Zagreva" to="Zagrijava" />
<Word from="zagrevanja" to="zagrijavanja" />
<Word from="zagrevanje" to="zagrijavanje" />
<Word from="Zagrevate" to="Zagrijavate" />
<Word from="zagrevanje" to="zagrijavanje" />
<Word from="zakunio" to="zakleo" />
@ -1316,9 +1327,11 @@
<Word from="Bejker" to="Baker" />
<Word from="Bitls" to="Beatles" />
<Word from="Bitlsi" to="Beatlesi" />
<Word from="Brejdi" to="Brady" />
<Word from="Bridžit" to="Bridget" />
<Word from="Brižit" to="Brigitte" />
<Word from="Bruks" to="Brooks" />
<Word from="Čipova" to="Chipova" />
<Word from="Dablin" to="Dublin" />
<Word from="Dablina" to="Dublina" />
<Word from="Dablinu" to="Dublinu" />
@ -1343,6 +1356,8 @@
<Word from="Džeri" to="Jerry" />
<Word from="Džerija" to="Jerryja" />
<Word from="Džo" to="Joe" />
<Word from="Džobs" to="Jobs" />
<Word from="Džobsu" to="Jobsu" />
<Word from="Džoel" to="Joel" />
<Word from="Džoelu" to="Joelu" />
<Word from="Džoš" to="Josh" />
@ -1360,6 +1375,7 @@
<Word from="Ebi" to="Abby" />
<Word from="Ejb" to="Abe" />
<Word from="Ejba" to="Abea" />
<Word from="Ejdrijen" to="Adrian" />
<Word from="Ejmi" to="Amy" />
<Word from="Ejpril" to="April" />
<Word from="Empajer" to="Empire" />
@ -1440,6 +1456,7 @@
<Word from="Mek" to="Mac" />
<Word from="Merilend" to="Maryland" />
<Word from="Mičel" to="Mitchell" />
<Word from="Mik" to="Mick" />
<Word from="Nensi" to="Nancy" />
<Word from="Niče" to="Nietzsche" />
<Word from="Ničea" to="Nietzschea" />
@ -1477,6 +1494,10 @@
<Word from="Šeron" to="Sharon" />
<Word from="Šeril" to="Cheryl" />
<Word from="Širli" to="Shirley" />
<Word from="Šon" to="Sean" />
<Word from="Šona" to="Seana" />
<Word from="Šone" to="Seane" />
<Word from="Šonu" to="Seanu" />
<Word from="Španiji" to="Španjolskoj" />
<Word from="Tajms" to="Times" />
<Word from="Tejlor" to="Taylor" />
@ -1535,6 +1556,8 @@
<LinePart from="Ako ej" to="Ako je" />
<LinePart from="ako ej" to="ako je" />
<LinePart from="bez svesti" to="bez svijesti" />
<LinePart from="Biće oni" to="Bit će oni" />
<LinePart from="biće ti žao" to="bit će ti žao" />
<LinePart from="Biću uz" to="Bit ću uz" />
<LinePart from="bi ja" to="bih ja" />
<LinePart from="bi la" to="bila" />
@ -1574,6 +1597,7 @@
<LinePart from="Izvini se" to="Ispričaj se" />
<LinePart from="izvini se" to="ispričaj se" />
<LinePart from="Izvinite me" to="Ispričajte me" />
<LinePart from="izvinite me" to="ispričajte me" />
<LinePart from="Izvinite nas" to="Ispričajte nas" />
<LinePart from="izvinite nas" to="ispričajte nas" />
<LinePart from="Izvinjavamo se" to="Ispričavamo se" />
@ -1603,7 +1627,9 @@
<LinePart from="mora da zna" to="mora znati" />
<LinePart from="moram da znam" to="moram znati" />
<LinePart from="Moram da znam" to="Moram znati" />
<LinePart from="Moraš da znaš" to="Moraš znati" />
<LinePart from="moraš da znaš" to="moraš znati" />
<LinePart from="Moraš da ideš" to="Moraš ići" />
<LinePart from="moraš da ideš" to="moraš ići" />
<LinePart from="može da bude" to="može biti" />
<LinePart from="možeš da budeš" to="možeš biti" />
@ -1620,6 +1646,7 @@
<LinePart from="neko nešto" to="netko nešto" />
<LinePart from="nedjelju dana" to="tjedan dana" />
<LinePart from="Ne mogu da verujem" to="Ne mogu vjerovati" />
<LinePart from="Nju Hempšir" to="New Hampshire" />
<LinePart from="new yorški" to="njujorški" />
<LinePart from="nju jorški" to="njujorški" />
<LinePart from="od kako" to="otkako" />
@ -2584,6 +2611,7 @@
<RegEx find="ćič" replaceWith="čić" />
<RegEx find="dilova" replaceWith="dila" />
<RegEx find="diteljk" replaceWith="diteljic" />
<RegEx find="dnstv" replaceWith="dinstv" />
<RegEx find="citov" replaceWith="citir" />
<RegEx find="cjenil" replaceWith="cijenil" />
<RegEx find="dnačin" replaceWith="dnadžb" />
@ -2666,6 +2694,7 @@
<RegEx find="m([io]t?)rić" replaceWith="m$1rit ć" />
<RegEx find="mi?j?enić" replaceWith="mijenit ć" />
<RegEx find="mjenim" replaceWith="mijenim" />
<RegEx find="mtić" replaceWith="mtit ć" />
<RegEx find="napredi" replaceWith="naprijedi" />
<RegEx find="ngažuj" replaceWith="ngažiraj" />
<RegEx find="nho" replaceWith="nko" />