Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-22 19:22:53 +01:00
Code Issues Projects Releases Wiki Activity

Update hrv_OCRFixReplaceList.xml

Browse Source
This commit is contained in:
Kruno H 2015-06-25 14:47:43 +02:00
parent 973b579384
commit 09e5512bc0
1 changed files with 29 additions and 47 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

76
Dictionaries/hrv_OCRFixReplaceList.xml
View File

@ -51,6 +51,7 @@
<Word from="Dali si" to="Da li si" />
<Word from="deda" to="djed" />
<Word from="dedom" to="djedom" />
<Word from="dejstvo" to="djelovanje" />
<Word from="deli" to="dijeli" />
<Word from="deluju" to="djeluju" />
<Word from="dete" to="dijete" />
@ -975,6 +976,7 @@
<RegEx find="(asvalt)(a|u|om|ira|irati|iranje|iran|iranog|irano|iranu|iranoj|iranom|na|ni|noj|nom)?" replaceWith="asfalt$2" />
<RegEx find="(avgust)(a|e|i|u|om|ima)?" replaceWith="kolovoz$2" />
<RegEx find="(avijon)(i|a|e|u|ima)" replaceWith="avion$2" />
<RegEx find="([bB])(ekstv)(a|u|om)" replaceWith="$1bijeg$3" />
<RegEx find="(bezbed)(an|na|no|ni|nima|nu|ne|nog|nom|noj)" replaceWith="sigur$2" />
<RegEx find="(bezbedno)(st|stima|šću|snu|snog|snom|sni)?" replaceWith="sigurno$2" />
<RegEx find="(bakcil)(i|a|e|u|ima)?" replaceWith="bacil$2" />
@ -985,28 +987,23 @@
<RegEx find="([bB])(eznadež)(an|na|no|nom|noj|nim|nima)" replaceWith="$1eznad$3" />
<RegEx find="([bB])(ež)(i|imo|e|ao|ala|ali|anje|anjem|anju)" replaceWith="$1jež$3" />
<RegEx find="([bB])(ijež)(i|imo|e|ao|ala|ali)" replaceWith="$1jež$3" />
<RegEx find="(bioskop)(a|u|om)" replaceWith="kin$2" />
<RegEx find="(b|B)(led)(a|e|i|o|u|im|om|ima)" replaceWith="$1lijed$3" />
<RegEx find="([bB])(ožiji)(a|i|u|om|ima)?" replaceWith="$1ožji$3" />
<RegEx find="(božić)(a|e|i|u|em|ima)?" replaceWith="Božić$2" />
<RegEx find="(Božićn)(a|e|i|o|u|om|im)?" replaceWith="božićn$2" />
<RegEx find="(Božičn)(a|e|i|o|u|om|im)?" replaceWith="božićn$2" />
<RegEx find="(bukvaln)(a|e|i|o|u|om|og|im|ima)?" replaceWith="doslovn$2" />
<RegEx find="(Hrist)(a|e|u|om)?" replaceWith="Krist$2" />
<RegEx find="(Californi)(a|ja|i|ji|u|ju)?" replaceWith="Kaliforni$2" />
<RegEx find="\b([cC])(j?el)(a|e|i|o|u|og|om|oj|ima|og|osti)\b" replaceWith="$1ijel$3" />
<RegEx find="\b([cC])(en)(a|e|i|o|u|om|ama|iti|ila|ilo|io|im|iš|imo|ite|iše)" replaceWith="$1ijen$3" />
<RegEx find="([cC])(enjen)(a|e|i|im|ima|o|u|om|oj)?" replaceWith="$1ijenjen$3" />
<RegEx find="([cC])(jenjen)(a|e|i|im|ima|o|u|om|oj)?" replaceWith="$1ijenjen$3" />
<RegEx find="([cC])(j?enjen)(a|e|i|im|ima|o|u|om|oj)?" replaceWith="$1ijenjen$3" />
<RegEx find="(čas)(a|u|om|ovi|ovima|ove|ova)" replaceWith="sat$2" />
<RegEx find="([čČ])(ove)(k|ka|ku|kom|kove|če|čanstvo|čanstva|čanstvu|čanstvom|čni|čna|čno)" replaceWith="$1ovje$3" />
<RegEx find="([čČ])(ovije)(k|ka|ku|kom|kov|kovog|kovoj|kovom|če|čanstva|čanstvu|čanstvom|čni|čna|čno)" replaceWith="$1ovje$3" />
<RegEx find="([bB])(ekstv)(a|u|om)" replaceWith="$1bijeg$3" />
<RegEx find="(bioskop)(a|u|om)" replaceWith="kin$2" />
<RegEx find="\b([cC])(el)(a|e|i|o|u|og|om|oj|ima|og|osti)\b" replaceWith="$1ijel$3" />
<RegEx find="\b([cC])(jel)(a|e|i|o|u|om|oj|ima|og|osti)\b" replaceWith="$1ijel$3" />
<RegEx find="(ćut)(im|iš|i|imo|ite|e)" replaceWith="šut$2" />
<RegEx find="(Ćut)(im|iš|i|imo|ite|e)" replaceWith="Šut$2" />
<RegEx find="(del)(ić|ića|iće|ićem|ići|ićima)" replaceWith="djel$2" />
<RegEx find="(del)(a|o|u)?" replaceWith="djel$2" />
<RegEx find="(del)(a|o|u|ić|ića|iće|ićem|ići|ićima)" replaceWith="djel$2" />
<RegEx find="\b([dD])(j?eli)(m|š|mo|te|li|le|la|ti)\b" replaceWith="$1ijel$3" />
<RegEx find="(delikvent)(a|e|i|u|ima|om|na|no|ni|nim|nom|noj|ica|ici|icu|ice|cija|ciji|ciju|cije|cijama)?" replaceWith="delinkvent$2" />
<RegEx find="([dD])(eluje)(m|š|mo|te|mo)?" replaceWith="$1jeluje$3" />
@ -1025,8 +1022,7 @@
<RegEx find="([dD])(obija)(m|š|mo|te|ju|la|le|li|ti)?" replaceWith="$1obiva$3" />
<RegEx find="(d|D)(oktork)(a|e|i|u|om|ama)" replaceWith="$1oktoric$3" />
<RegEx find="(dopada)(m|š|mo|te|ju|la|le|li|ti)?" replaceWith="sviđa$2" />
<RegEx find="(doprine)(te|la|le|li|lo|t|ti|to|vši)" replaceWith="doprinije$2" />
<RegEx find="(doprinje)(te|la|le|li|lo|t|ti|to|vši)" replaceWith="doprinije$2" />
<RegEx find="(doprinj?e)(te|la|le|li|lo|t|ti|to|vši)" replaceWith="doprinije$2" />
<RegEx find="(doprin)(eo|eso|jeo)?" replaceWith="doprinio" />
<RegEx find="(dosije)(a|e|i|u|ima)" replaceWith="dosje$2" />
<RegEx find="(dospe)(li|la|ti)" replaceWith="dospje$2" />
@ -1043,14 +1039,13 @@
<RegEx find="(Fudbal)(a|u|om)?" replaceWith="Nogomet$2" />
<RegEx find="(Frojd)(a|e|u|om|ovom|ovoj|ovim)?" replaceWith="Freud$2" />
<RegEx find="([gG])(luv)(a|e|o|u|oj|om)?" replaceWith="$1luh$3" />
<RegEx find="\b([gG])(reh)(a|e|u|om)?" replaceWith="$1rijeh$3" />
<RegEx find="\b([gG])(rjeh)(a|e|u|om)?" replaceWith="$1rijeh$3" />
<RegEx find="\b([gG])(rj?eh)(a|e|u|om)?" replaceWith="$1rijeh$3" />
<RegEx find="\b([gG])(res)(i|ima)" replaceWith="$1rijes$3" />
<RegEx find="([gG])(rejanj)(e|u|a|em|ima)" replaceWith="$1rijanj$3" />
<RegEx find="([gG])(reši)(o|m|š|mo|te|la|li|ti)" replaceWith="$1riješi$3" />
<RegEx find="([gG])(rješi)(o|m|š|mo|te|la|li|ti)" replaceWith="$1riješi$3" />
<RegEx find="([gG])(rj?eši)(o|m|š|mo|te|la|li|ti)" replaceWith="$1riješi$3" />
<RegEx find="(haos)(a|i|e|u|ima|na|no|ni|om)?" replaceWith="kaos$2" />
<RegEx find="(h|H)(emij)(a|e|i|ski|ska|sko|skom|skoj|skih)" replaceWith="$1emij$3" />
<RegEx find="(Hrist)(a|e|u|om)?" replaceWith="Krist$2" />
<RegEx find="(hrišćan)(a|e|i|u|in|ina|ine|inu|inom|ka|ku|ki|kama|ima|ska|ske|ski|sko|sku|skoj|skom)" replaceWith="kršćan$2" />
<RegEx find="(Holand)(ska|sku|skom|skoj)" replaceWith="Nizozems$3" />
<RegEx find="(Nizuzem)(ska|sku|skom|skoj)" replaceWith="Nizozems$3" />
@ -1101,26 +1096,19 @@
<RegEx find="(k|K)(uvan)(a|e|i|o|u|ima|og|om|oj|im)?" replaceWith="$1uhan$3" />
<RegEx find="(k|K)(uva)(r|re|ri|ru|rima|rica|ricu|rice|m|š|mo|te|ju|še|li|ti)" replaceWith="$1uha$3" />
<RegEx find="\b(krst)(a|u|em)?" replaceWith="križ$2" />
<RegEx find="([lL])(eči)(o|la|le|li|ti|še)?" replaceWith="$1iječi$3" />
<RegEx find="([lL])(ječi)(o|la|le|li|ti|še)?" replaceWith="$1iječi$3" />
<RegEx find="([lL])(ečni)(k|ka|ku|ca|cu|ci|ce|cima|cama)?" replaceWith="$1iječni$3" />
<RegEx find="([lL])(ječni)(k|ka|ku|ca|cu|ci|ce|cima|cama)?" replaceWith="$1iječni$3" />
<RegEx find="([lL])(j?eči)(o|la|le|li|ti|še)?" replaceWith="$1iječi$3" />
<RegEx find="([lL])(j?ečni)(k|ka|ku|ca|cu|ci|ce|cima|cama)" replaceWith="$1iječni$3" />
<RegEx find="([lL])(ekar)(a|e|u|om)?" replaceWith="$1iječnik$3" />
<RegEx find="(l|L)(ekov)(a|e|i|ima)" replaceWith="$1ijekov$3" />
<RegEx find="(l|L)(jekov)(a|e|i|ima)" replaceWith="$1ijekov$3" />
<RegEx find="(l|L)(ečenj)(a|e|u|em|ima)" replaceWith="$1iječenj$3" />
<RegEx find="(l|L)(ječenj)(a|e|u|em|ima)" replaceWith="$1iječenj$3" />
<RegEx find="(l|L)(ečen)(a|e|o|u|om|og|ima)" replaceWith="$1iječen$3" />
<RegEx find="(l|L)(ječen)(a|e|o|u|om|og|ima)" replaceWith="$1iječen$3" />
<RegEx find="\b(len)(a|e|i|o|u|om|ima|čina|čino|čine)" replaceWith="lijen$2" />
<RegEx find="\b(Len)(a|e|i|o|u|om|ima|čina|čino|čine)" replaceWith="Lijen$2" />
<RegEx find="(l|L)(j?ekov)(a|e|i|ima)" replaceWith="$1ijekov$3" />
<RegEx find="(l|L)(j?ečenj)(a|e|u|em|ima)" replaceWith="$1iječenj$/>
<RegEx find="(l|L)(j?ečen)(a|e|o|u|om|og|ima)?" replaceWith="$1iječen$3" />
<RegEx find="\b(l|L)(en)(a|e|i|o|u|om|ima|čina|čino|čine)?" replaceWith="lijen$2" />
<RegEx find="([lL])(ep)(a|i|o|u|e|im|om|ima|og|om)?\b" replaceWith="$1ijep$3" />
<RegEx find="([lL])(epš)(a|i|o|u|e|im|om|ima|og|om)?\b" replaceWith="$1jepš$3" />
<RegEx find="\b(lev)(a|i|o|u|om|oj|ima)" replaceWith="lijev$2" />
<RegEx find="\b(ličn)(a|e|i|o|u|im|om|oj)?" replaceWith="osobn$2" />
<RegEx find="\b(Ličn)(a|e|i|o|u|im|om|oj)?" replaceWith="Osobn$2" />
<RegEx find="(lobanj)(a|e|i|u|om|aima)?" replaceWith="lubanj$2" />
<RegEx find="(Lobanj)(a|e|i|u|om|aima)?" replaceWith="Lubanj$2" />
<RegEx find="(l|L)(obanj)(a|e|i|u|om|ama)?" replaceWith="$1ubanj$3" />
<RegEx find="\b(ljep)(a|e|i|o|u|om|oj|ima)\b" replaceWith="lijep$2" />
<RegEx find="\b(Ljep)(a|e|i|o|u|om|oj|ima)\b" replaceWith="Lijep$2" />
<RegEx find="(l|L)(uda)(k|ci|čka|čki|čko|čkom|čku|kinja|kinju|kinje)?" replaceWith="$1uđa$3" />
@ -1131,15 +1119,13 @@
<RegEx find="([mM])(j?enja)(m|š|mo|te|ju|li|ti)?" replaceWith="$1ijenja$3" />
<RegEx find="([mM])(est)(a|o|u|om|ima)" replaceWith="$1jest$3" />
<RegEx find="([mM])(jenja)(m|š|mo|te|ju|li)?" replaceWith="$1ijenja$3" />
<RegEx find="([mM])(lek)(a|o|u|om)" replaceWith="$1lijek$3" />
<RegEx find="([mM])(ljek)(a|o|u|om)" replaceWith="$1lijek$3" />
<RegEx find="([mM])(lj?ek)(a|o|u|om)" replaceWith="$1lijek$3" />
<RegEx find="([mM])(oč)(i|u|ni|nim|noj)" replaceWith="$1oć$3" />
<RegEx find="([mM])(ogučnost)(i|ima)?" replaceWith="$1ogućnost$3" />
<RegEx find="([mM])(uva)(o|j|š|mo|te|ju|vši|li|ti|lo|la|le)" replaceWith="$1ota$3" />
<RegEx find="(muzik)(a|u|om)\b" replaceWith="glazb$2" />
<RegEx find="([nN])(amer)(a|e|i|u|om|na|no|noj|nom|nim)" replaceWith="$1amjer$3" />
<RegEx find="([nN])(ameni)(m|š|mo|o|te|ti|o|la|le|li|še)?" replaceWith="$1amijeni$3" />
<RegEx find="([nN])(amjeni)(m|š|mo|o|te|ti|o|la|le|li|še)?" replaceWith="$1amijeni$3" />
<RegEx find="([nN])(amj?eni)(m|š|mo|o|te|ti|o|la|le|li|še)?" replaceWith="$1amijeni$3" />
<RegEx find="([nN])(amesti)(m|š|mo|te|ti|o|la|le|li|še)?" replaceWith="$1amjesti$3" />
<RegEx find="(n|N)(aočar)(a|e|ama|ima)" replaceWith="$1aočal$3" />
<RegEx find="([nN])(arandž)(a|e|i|u|om|ama)" replaceWith="$aranč$3" />
@ -1148,15 +1134,14 @@
<RegEx find="([nN])(asmej)(em|eš|e|emo|ete|u|ani|ane|anima|anoj|anom|ano|ava|avaš|an|ana|anu|anog|alo|ala|ao)" replaceWith="$1asmij$3" />
<RegEx find="([nN])(ater)(a|ale|ali|ala|alo|an|ana|anoj|anom|anim|ati|alo|ao)" replaceWith="$1atjer$3" />
<RegEx find="([nN])(ečove)(k|ka|ku|kom|kove|če|čanstvo|čanstva|čanstvu|čanstvom|čni|čna|čno)" replaceWith="$1ečovje$3" />
<!-- vrijedi i za ponedjeljak -->
<RegEx find="(nedelj)(a)?" replaceWith="nedjelj$2" />
<!-- vrijedi i za ponedjeljak* -->
<RegEx find="(nedelj)(a)*" replaceWith="nedjelj$2" />
<RegEx find="(n|N)(emačk)(a|e|i|o|u|om|im)" replaceWith="$1jemačk$3" />
<RegEx find="([nN])(eprijatn)(a|e|i|u|om|no|noj|nom|nim)?" replaceWith="$1eugodn$3" />
<RegEx find="([nN])(eprimet)(an|na|no|ne|ljiv|ljiva|ljivo|ljivost|ljivošću|ljivosti|nost)" replaceWith="$1eprimjet$3" />
<RegEx find="([nN])(epobediv)(a|o|om|oj|e|i|u|ost|ošću|osti)" replaceWith="$1epobjediv$3" />
<RegEx find="([nN])(ereš)(en|eno|ena|ene|eni)" replaceWith="$1eriješ$3" />
<RegEx find="(nevaspitan)(a|e|i|o|u|oj|om|im|ima|je|ju|jem)" replaceWith="neobrazovan$2" />
<RegEx find="(nevaspitan)(a|e|i|o|u|oj|om|im|ima|je|ju|jem)" replaceWith="Neobrazovan$2" />
<RegEx find="(n|N)(evaspitan)(a|e|i|o|u|oj|om|im|ima|je|ju|jem)?" replaceWith="$1eobrazovan$3" />
<!-- vrijedi i za vjerojatno -->
<RegEx find="([nN])(everojat)(an|na|ne|ni|no|nom|nog|nu|noj|nim)*" replaceWith="$1evjerojat$3" />
<!-- -||- -->
@ -1166,8 +1151,7 @@
<RegEx find="([nN])(ežn)(a|e|i|u|om|og|oj|ima|iji|ije|ija)" replaceWith="$1ježn$3" />
<RegEx find="([nN])(oč)(i|u|ni|na|noj|nim)?" replaceWith="$1oć$3" />
<RegEx find="(naučn)(i|a|o|og|u|oj|om|ik|im|ici)?" replaceWith="znanstven$2" />
<RegEx find="([oO])(bavest)(io|ila|ile|ili|iti|iše)?" replaceWith="$1bavijest$3" />
<RegEx find="([oO])(bavjest)(io|ila|ile|ili|iti|iše)?" replaceWith="$1bavijest$3" />
<RegEx find="([oO])(bavj?est)(io|ila|ile|ili|iti|iše|ima|i)?" replaceWith="$1bavijest$3" />
<RegEx find="(obezbeđenj)(a|e|u|ima)" replaceWith="osiguranj$2" />
<RegEx find="([oO])(brača)(m|o|š|mo|te|ju|jući|la|le|li|lo|ti|vši)?" replaceWith="$1braća$3" />
<RegEx find="\b(odeć)(a|e|u|i|om)" replaceWith="odjeć$2" />
@ -1184,8 +1168,7 @@
<RegEx find="([oO])(seti)(o|m|š|la|li|le|ti)" replaceWith="$1sjeti$3" />
<RegEx find="([oO])(setljiv)(a|i|e|o|om|im|ima|oj)?" replaceWith="$1sjetljiv$3" />
<RegEx find="([oO])(seća)(m|š|mo|te|ju|j|ja|ji|je|jem|ju|jima)?" replaceWith="$1sjeća$3" />
<RegEx find="([oO])(seča)(m|š|mo|te|ju|j|ja|ji|je|jem|ju|jima)?" replaceWith="$1sjeća$3" />
<RegEx find="([oO])(sječa)(m|š|mo|te|ju|j|ja|ji|je|jem|ju|jima)?" replaceWith="$1sjeća$3" />
<RegEx find="([oO])(sj?eča)(m|š|mo|te|ju|j|ja|ji|je|jem|ju|jima)?" replaceWith="$1sjeća$3" />
<RegEx find="([oO])(strv)(a|u|om)" replaceWith="$1tok$3" />
<RegEx find="([oO])(sveženj)(a|e|u|em|ima)" replaceWith="$1svježenj$3" />
<RegEx find="([oO])(svežava)(m|š|mo|te|ju|ti|li|la|lo|ju|juć|juće|juća)?" replaceWith="$1svježava$3" />
@ -1239,11 +1222,9 @@
<RegEx find="\b([pP])(obed)(im|iš|i|imo|ite|e|iti|ili|ivši)" replaceWith="$1obijed$3" />
<RegEx find="([pP])(obeg)(ao|la|li|le|lo|avši)" replaceWith="$1objeg$3" />
<RegEx find="([pP])(oent)(a|e|u|i|o|om|ama)" replaceWith="$1oant$3" />
<RegEx find="([pP])(ogreš)(io|ila|ili)" replaceWith="$1ogriješ$3" />
<RegEx find="([pP])(ogrješ)(io|ila|ili)" replaceWith="$1ogriješ$3" />
<RegEx find="([pP])(ogrj?eš)(io|ila|ili)" replaceWith="$1ogriješ$3" />
<RegEx find="([pP])(oslednj)(i|im|e|em|eg|ima)" replaceWith="$1osljed$3" />
<RegEx find="(pomera)(m|š|mo|te)" replaceWith="miče$2" />
<RegEx find="([pP])(onj?e)(l|t)" replaceWith="$1onije$3" />
<RegEx find="([pP])(overenj)(e|a|u|em)" replaceWith="$1ovjerenj$3" />
<RegEx find="([pP])(overljiv)(a|e|i|o|u|ima|ima|om|og|oj)?" replaceWith="$1ovjerljiv$3" />
<RegEx find="\b([pP])(revoz)(a|u|om|ovi|ovima)?\b" replaceWith="$1rijevoz$3" />
@ -1266,7 +1247,7 @@
<RegEx find="(prevazi)(đen|đeno|đena|ći|laženje|laženja|laženju|laženjem)" replaceWith="nadi$2" />
<RegEx find="([rR])(azmen)(a|e|u|i|ama)?" replaceWith="$1azmjen$3" />
<RegEx find="([rR])(azume)(m|š|mo|te|va)" replaceWith="$1azumije$3" />
<RegEx find="(redhodn)" replaceWith="rethodn" />
<RegEx find="(redhodn)" replaceWith="rethodn" />
<RegEx find="([rR])(eligijozn)(a|e|i|o|u|om|im|ima)" replaceWith="$1eligiozn$3" />
<RegEx find="([rR])(ešava)(m|š|mo|te|ju|nje|li)?" replaceWith="$1ješava$3" />
<RegEx find="([rR])(ešenj)(e|em|u|a|ima)?" replaceWith="$1ješenj$3" />
@ -1424,12 +1405,13 @@
<RegEx find="([žŽ])(lezd)(a|e|i|o|u|ama)" replaceWith="$1lijezd$3" />
<RegEx find="([žŽ])(ljezd)(a|e|i|o|u|ama)" replaceWith="$1lijezd$3" />
<RegEx find="(žurk)(a|e|i|u|om|ama)" replaceWith="zabav$2" />
<RegEx find="(z|Z)(amnom)" replaceWith="$1a mnom" />
<RegEx find="(s|S)(amnom)" replaceWith="$1a mnom" />
<RegEx find="(s|S)(amnom)" replaceWith="$1a mnom" />
<!-- experimental -->
<RegEx find="(vređ)(a)*" replaceWith="vrijeđ$2" />
<RegEx find="(oćeju)" replaceWith="oće" />
<RegEx find="(tćeš)" replaceWith="t ćeš" />
<RegEx find="(zovati)" replaceWith="zirati" />
</RegularExpressions>
</OCRFixReplaceList>
</OCRFixReplaceList>