Merge pull request #2726 from diomed/patch-1

Croatian OCR Fixes

Dictionaries/hrv_OCRFixReplaceList.xml

@@ -62,6 +62,8 @@
     <Word from="cevi" to="cijevi" />
     <Word from="cevčica" to="cjevčica" />
     <Word from="cevčicu" to="cjevčicu" />
+    <Word from="cvetaju" to="cvjetaju" />
+    <Word from="cvetove" to="cvjetove" />
     <Word from="cvetu" to="cvijetu" />
     <Word from="cvjetu" to="cvijetu" />
     <Word from="cvetalo" to="cvjetalo" />
@@ -159,6 +161,7 @@
     <Word from="emituje" to="emitira" />
     <Word from="emitujem" to="emitiram" />
     <Word from="emituju" to="emitiraju" />
+    <Word from="evra" to="eura" />
     <Word from="familijama" to="obiteljima" />
     <Word from="familiji" to="obitelji" />
     <Word from="flertuje" to="očijuka" />
@@ -204,8 +207,6 @@
     <Word from="Ignorišite" to="Ignorirajte" />
     <Word from="ignorišu" to="ignoriraju" />
     <Word from="ilustruju" to="ilustriraju" />
-    <Word from="informiše" to="informira" />
-    <Word from="informišu" to="informiraju" />
     <Word from="inspirisao" to="inspirirao" />
     <Word from="interesuje" to="zanima" />
     <Word from="Interesuje" to="Zanima" />
@@ -221,13 +222,16 @@
     <Word from="ivice" to="ruba" />
     <Word from="ivici" to="rubu" />
     <Word from="ivicu" to="rub" />
+    <Word from="izoluje" to="izolira" />
+    <Word from="izoluješ" to="izoliraš" />
     <Word from="Izolujete" to="Izolirate" />
     <Word from="Izolujte" to="Izolirajte" />
+    <Word from="izolujte" to="izolirajte" />
     <Word from="izgladneo" to="izgladnio" />
     <Word from="izmeriti" to="izmjeriti" />
     <Word from="izmerio" to="izmjerio" />
+    <Word from="izmešane" to="izmiješane" />
     <Word from="izneo" to="iznio" />
-    <Word from="izolujte" to="izolirajte" />
     <Word from="izneti" to="iznijeti" />
     <Word from="izvestan" to="izvjestan" />
     <Word from="izvinim" to="ispričam" />
@@ -239,6 +243,10 @@
     <Word from="jedamput" to="jedanput" />
     <Word from="jelda" to="jel' da" />
     <Word from="Ješćemo" to="Jest ćemo" />
+    <Word from="ješće" to="jest će" />
+    <Word from="ješću" to="jest ću" />
+    <Word from="ješćeš" to="jest ćeš" />
+    <Word from="ješćemo" to="jest ćemo" />
     <Word from="juče" to="jučer" />
     <Word from="Juče" to="Jučer" />
     <Word from="kakava" to="kakva" />
@@ -336,6 +344,10 @@
     <Word from="leve" to="lijeve" />
     <Word from="loži" to="pali" />
     <Word from="lza" to="Iza" />
+    <Word from="ljepiti" to="lijepiti" />
+    <Word from="ljepili" to="lijepili" />
+    <Word from="magnezijuma" to="magnezija" />
+    <Word from="magnezijumu" to="magneziju" />
     <Word from="majca" to="majica" />
     <Word from="majce" to="majice" />
     <Word from="majcu" to="majicu" />
@@ -589,6 +601,8 @@
     <Word from="praktikovati" to="prakticirati" />
     <Word from="pre" to="prije" />
     <Word from="Pre" to="Prije" />
+    <Word from="predela" to="predjela" />
+    <Word from="predelu" to="predjelu" />
     <Word from="Preći" to="Prijeći" />
     <Word from="preći" to="prijeći" />
     <Word from="prećutkuje" to="prešućuje" />
@@ -629,6 +643,8 @@
     <Word from="prmda" to="iako" />
     <Word from="procenat" to="postotak" />
     <Word from="procent" to="postotak" />
+    <Word from="procenta" to="postotka" />
+    <Word from="procentu" to="postotku" />
     <Word from="procenata" to="postotaka" />
     <Word from="procenti" to="postoci" />
     <Word from="Procenti" to="Postoci" />
@@ -649,6 +665,7 @@
     <Word from="protestvuju" to="protestiraju" />
     <Word from="protestujući" to="protestirajući" />
     <Word from="psihićki" to="psihički" />
+    <Word from="pto" to="što" />
     <Word from="ptom" to="potom" />
     <Word from="računari" to="računala" />
     <Word from="računare" to="računala" />
@@ -677,6 +694,8 @@
     <Word from="Resetujte" to="Resetirajte" />
     <Word from="rijeđi" to="rjeđi" />
     <Word from="Rizikuj" to="Riskiraj" />
+    <Word from="rizikuju" to="riskiraju" />
+    <Word from="rizikuje" to="riskira" />
     <Word from="rokenrol" to="rock'n'roll" />
     <Word from="ronioc" to="ronilac" />
     <Word from="saviješću" to="savješću" />
@@ -706,6 +725,7 @@
     <Word from="sigruno" to="sigurno" />
     <Word from="siguan" to="siguran" />
     <Word from="sija" to="sja" />
+    <Word from="sijaju" to="sjaju" />
     <Word from="sirće" to="ocat" />
     <Word from="sirćeta" to="octa" />
     <Word from="sirćetu" to="octu" />
@@ -788,6 +808,7 @@
     <Word from="svjesti" to="svijesti" />
     <Word from="Svjesti" to="Svijesti" />
     <Word from="svetova" to="svjetova" />
+    <Word from="svetove" to="svjetove" />
     <Word from="svugde" to="svugdje" />
     <Word from="šolja" to="šalica" />
     <Word from="šolju" to="šalicu" />
@@ -943,6 +964,7 @@
     <Word from="vrteo" to="vrtio" />
     <Word from="whiskey" to="viski" />
     <Word from="zaceljivanje" to="zacjeljivanje" />
+    <Word from="zagreva" to="zagrijava" />
     <Word from="zagrevanje" to="zagrijavanje" />
     <Word from="zalepili" to="zalijepili" />
     <Word from="zalepiti" to="zalijepiti" />
@@ -969,12 +991,17 @@
     <Word from="zvaničan" to="služben" />
     <Word from="žemo" to="ćemo" />
     <Word from="Živeo" to="Živio" />
+    <Word from="živeo" to="živio" />
     <Word from="žmureo" to="žmirio" />
 
     <!-- imena -->
     <Word from="Alis" to="Alice" />
     <Word from="Ajron" to="Iron" />
     <Word from="Ajvi" to="Ivy" />
+    <Word from="Anđeles" to="Angeles" />
+    <Word from="Anđelesa" to="Angelesa" />
+    <Word from="Anđelesu" to="Angelesu" />
+    <Word from="Anđelesom" to="Angelesom" />
     <Word from="Arči" to="Archie" />
     <Word from="Beverli" to="Beverly" />
     <Word from="Bejker" to="Baker" />
@@ -1005,6 +1032,7 @@
     <Word from="Džejn" to="Jane" />
     <Word from="Džil" to="Jill" />
     <Word from="Džodi" to="Jody" />
+    <Word from="Džud" to="Jude" />
     <Word from="Džuli" to="Julie" />
     <Word from="Ebi" to="Abby" />
     <Word from="Ejb" to="Abe" />
@@ -1043,7 +1071,7 @@
     <Word from="Klod" to="Claude" />
     <Word from="Klode" to="Claude" />
     <Word from="Kortni" to="Courtney" />
-    <Word from="Krejg" to="Craig" />
+    <Word from="Krej?g" to="Craig" />
     <Word from="Kristi" to="Christie" />
     <Word from="Lajf" to="Life" />
     <Word from="Lejk" to="Lake" />
@@ -1067,10 +1095,15 @@
     <Word from="Odri" to="Audrey" />
     <Word from="Ohajo" to="Ohio" />
     <Word from="Ohaja" to="Ohia" />
+    <Word from="Ohaju" to="Ohiu" />
+    <Word from="Orlins" to="Orleans" />
+    <Word from="Orlinsa" to="Orleansa" />
+    <Word from="Orlinsu" to="Orleansu" />
     <Word from="Pem" to="Pam" />
     <Word from="Pejdž" to="Paige" />
     <Word from="Rejčel" to="Rachel" />
     <Word from="Rouz" to="Rose" />
+    <Word from="Sidnej" to="Sidney" />
     <Word from="Sole" to="Saule" />
     <Word from="Stejt" to="State" />
     <Word from="Stejsi" to="Stacy" />
@@ -1143,6 +1176,7 @@
     <LinePart from="dali ću" to="hoću li" />
     <LinePart from="Da li je" to="Je li" />
     <LinePart from="da li je" to="je li" />
+    <LinePart from="dali je" to="je li" />
     <LinePart from="Da li ste" to="Jeste li" />
     <LinePart from="Da li si" to="Jesi li" />
     <LinePart from="dali si" to="jesi li" />
@@ -1186,6 +1220,7 @@
     <LinePart from="Moramo da idemo" to="Moramo ići" />
     <LinePart from="moraš da ideš" to="moraš ići" />
     <LinePart from="može da bude" to="može biti" />
+    <LinePart from="može da diše" to="može disati" />
     <LinePart from="možemo da imamo" to="možemo imati" />
     <LinePart from="na večer" to="navečer" />
     <LinePart from="Na večer" to="Navečer" />
@@ -1227,6 +1262,7 @@
     <LinePart from="želim da budem" to="želim biti" />
     <LinePart from="Želim da budem" to="Želim biti" />
     <LinePart from="želiš da budeš" to="želiš biti" />
+    <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="želim da umrem" to="želim umrijeti" />
@@ -1643,7 +1679,7 @@
     <RegEx find="esmij" replaceWith="e smij" />
     <RegEx find="esre[cćč]n" replaceWith="esretn" />
     <RegEx find="esvj?est" replaceWith="esvijest" />
-    <RegEx find="(?&lt;!hv)ališ[eu]" replaceWith="alizira" />
+    <RegEx find="(?&lt;!mhv)ališ[eu]" replaceWith="alizira" />
     <RegEx find="([nN])ezna([mš]|mo|t[ei]|ju|jući)?\b" replaceWith="$1e zna$2" />
     <RegEx find="([nN])ežn" replaceWith="$1ježn" />
     <RegEx find="\b([nN])oč([iu]|n[aeiou]|no[mjg]|nim)?" replaceWith="$1oć$2" />
@@ -2134,6 +2170,7 @@
     <RegEx find="adji" replaceWith="ađi" />
     <RegEx find="(?&lt;![ml])a([blcrnz])ić" replaceWith="a$1it ć" />
     <RegEx find="a(jm|mn)(t?)ić" replaceWith="a$1$2it ć" />
+    <RegEx find="ajuc" replaceWith="ajuć" />
     <RegEx find="\bajpod" replaceWith="iPod" />
     <RegEx find="alepn" replaceWith="aljepn" />
     <RegEx find="alj?epi" replaceWith="alijepi" />
@@ -2191,13 +2228,15 @@
     <RegEx find="(mpl|st|f)ikova([nlot])" replaceWith="$1icira$2" />
     <RegEx find="fikuj" replaceWith="ficir" />
     <RegEx find="(fin|form)isa" replaceWith="$1ira" />
+    <RegEx find="(?&lt;o)formiš" replaceWith="formiraj" />
     <RegEx find="fi(sa|še)" replaceWith="fira" />
     <RegEx find="frov" replaceWith="frir" />
     <RegEx find="grišć" replaceWith="grist ć" />
+    <RegEx find="haot" replaceWith="kaot" />
     <RegEx find="hrono" replaceWith="krono" />
     <RegEx find="(?&lt;!M)ick([eoiu])" replaceWith="ičk$1" />
     <RegEx find="idev" replaceWith="idjev" />
-    <RegEx find="ijoz" replaceWith="ioz" />
+    <RegEx find="ijo([absdegkptz])" replaceWith="io$1" />
     <RegEx find="interesuje" replaceWith="zanima" />
     <RegEx find="([ai])nić" replaceWith="$1nit ć" />
     <RegEx find="(?&lt;!\b[Oo]zlo|\b[Ii]sp(rip)?ov|i)jeđen" replaceWith="ijeđen" />
@@ -2212,6 +2251,7 @@
     <!-- doktorica / profesorica -->
     <RegEx find="(kt|s)ork" replaceWith="$1oric" />
     <RegEx find="ktovanj" replaceWith="ktiranj" />
+    <RegEx find="ktuj" replaceWith="ktiraj" />
     <RegEx find="(?&lt;![Oo]b)([nl])izuj[eu]" replaceWith="$1izira" />
     <RegEx find="\bl([čćžzvsmrtpdbgkn])" replaceWith="i$1" />
     <RegEx find="\bL([čćžzvsmrtpdbgkn])" replaceWith="I$1" />
@@ -2266,6 +2306,7 @@
     <RegEx find="orks" replaceWith="orsk" />
     <RegEx find="osejan" replaceWith="osijan" />
     <RegEx find="o([dbjpv])ać(?!o)" replaceWith="o$1at ć" />
+    <RegEx find="oslj?ep" replaceWith="oslijep" />
     <RegEx find="sj?etić" replaceWith="sjetit ć" />
     <RegEx find="osb" replaceWith="osob" />
     <RegEx find="ovinova([lt])" replaceWith="okorava$1" />
@@ -2277,6 +2318,7 @@
     <RegEx find="(?&lt;![Pp]r|[Nn])adje(?!(v|n(e|u[olt]))\b)" replaceWith="ađe" />
     <RegEx find="par nedj?elja" replaceWith="par tjedana" />
     <RegEx find="pendov" replaceWith="pendir" />
+    <RegEx find="ralijs" replaceWith="rals" />
     <RegEx find="rasčist" replaceWith="raščist" />
     <RegEx find="kratij" replaceWith="kracij" />
     <RegEx find="(?&lt;![is]|S)redić" replaceWith="rjedit ć" />
@@ -2309,6 +2351,7 @@
     <RegEx find="stvrn" replaceWith="stvarn" />
     <RegEx find="(?&lt;!n)svesti([lt])" replaceWith="svijest$1" />
     <RegEx find="svetić" replaceWith="svetit ć" />
+    <RegEx find="([šđ])ać" replaceWith="$1ač" />
     <RegEx find="šević" replaceWith="ševit ć" />
     <RegEx find="šunjać" replaceWith="šuljat ć" />
     <RegEx find="tarisa" replaceWith="tira" />
@@ -2345,6 +2388,7 @@
     <RegEx find="umere" replaceWith="umjere" />
     <RegEx find="upoce" replaceWith="upocje" />
     <RegEx find="urfovan" replaceWith="urfan" />
+    <RegEx find="ušać" replaceWith="ušat ć" />
     <RegEx find="(?&lt;!kr|m)užić" replaceWith="užit ć" />
     <RegEx find="veri([lt])" replaceWith="vjeri$1" />
     <RegEx find="([do])vešć" replaceWith="$1vest ć" />
@@ -2376,13 +2420,16 @@
     <RegEx find="rčer" replaceWith="rcher" />
     <RegEx find="Avgani" replaceWith="Afgani" />
     <RegEx find="Belgijan" replaceWith="Belgij" />
+    <RegEx find="Bendžamin" replaceWith="Benjamin" />
     <RegEx find="Betmen" replaceWith="Batman" />
+    <RegEx find="Blejk" replaceWith="Blake" />
     <RegEx find="Bruklin" replaceWith="Brooklyn" />
     <RegEx find="Brajan" replaceWith="Brian" />
     <RegEx find="Brajs" replaceWith="Bryce" />
     <RegEx find="Braun" replaceWith="Brown" />
     <RegEx find="Brussels" replaceWith="Bruxelles" />
     <RegEx find="Čarls" replaceWith="Charles" />
+    <RegEx find="Čerčil" replaceWith="Churchill" />
     <RegEx find="Čikag" replaceWith="Chicag" />
     <RegEx find="Dajan([aeiou]|om)" replaceWith="Dian$1" />
     <RegEx find="Dankan" replaceWith="Duncan" />
@@ -2441,7 +2488,6 @@
     <RegEx find="livlend" replaceWith="leveland" />
     <RegEx find="(?&lt;![LT])ejl" replaceWith="ale" />
     <RegEx find="Losanđeles" replaceWith="Los Angeles" />
-    <RegEx find="Los Anđeles" replaceWith="Los Angeles" />
     <RegEx find="Majami" replaceWith="Miami" />
     <RegEx find="Majkl" replaceWith="Michael" />
     <RegEx find="Marfi" replaceWith="Murphy" />