dictionaries: automated XML upkeep

Dictionaries/bos_OCRFixReplaceList.xml

@@ -1,4 +1,4 @@
-<OCRFixReplaceList>
+<OCRFixReplaceList>
   <WholeWords>
     <Word from="andele" to="anđele" />
     <Word from="andeli" to="anđeli" />
@@ -247,7 +247,7 @@
     <Word from="glupaco" to="glupačo" />
     <Word from="govoris" to="govoriš" />
     <Word from="gradani" to="građani" />
-		<Word from="gradanin" to="građanin"/>
+    <Word from="gradanin" to="građanin" />
     <Word from="gradanine" to="građanine" />
     <Word from="gradic" to="gradić" />
     <Word from="gradica" to="gradića" />
@@ -1241,7 +1241,7 @@
     <LinePart from="Na vecer" to="Navečer" />
     <LinePart from="ne cu" to="neću" />
     <LinePart from="ne ces" to="nećeš" />
-		<LinePart  from="nešto sto"  to="nešto što"/>
+    <LinePart from="nešto sto" to="nešto što" />
     <LinePart from="ono sto" to="ono što" />
     <LinePart from="Ono sto" to="Ono što" />
     <LinePart from="reci ću" to="reći ću" />

Dictionaries/da_NoBreakAfterList.xml

@@ -1,12 +1,11 @@
 <NoBreakAfterList>
-  <Item>Dr.</Item>
   <Item>Dr</Item>
-  <Item>Hr.</Item>
-  <Item>Mr.</Item>
-  <Item>Mr</Item>
-  <Item>Mrs.</Item>
-  <Item>Ms.</Item>
-  <Item>Frk.</Item>
+  <Item>Dr.</Item>
   <Item>Frk.</Item>
   <Item>Fru</Item>
+  <Item>Hr.</Item>
+  <Item>Mr</Item>
+  <Item>Mr.</Item>
+  <Item>Mrs.</Item>
+  <Item>Ms.</Item>
 </NoBreakAfterList>

Dictionaries/en_NoBreakAfterList.xml

@@ -1,7 +1,7 @@
 <NoBreakAfterList>
+  <Item>Dr</Item>
   <Item>Dr.</Item>
   <Item>Mr.</Item>
   <Item>Mrs.</Item>
   <Item>Ms.</Item>
-  <Item>Dr</Item>
 </NoBreakAfterList>

Dictionaries/en_US_user.xml

@@ -24,6 +24,7 @@
   <word>arsenic</word>
   <word>arses</word>
   <word>astatine</word>
+  <word>astrometry</word>
   <word>attosecond</word>
   <word>autotrophs</word>
   <word>axe</word>
@@ -107,6 +108,7 @@
   <word>endeth</word>
   <word>episiotomy</word>
   <word>erbium</word>
+  <word>esophaguses</word>
   <word>euro</word>
   <word>europium</word>
   <word>euros</word>
@@ -205,6 +207,7 @@
   <word>mercury</word>
   <word>mesenteric</word>
   <word>metres</word>
+  <word>micrografting</word>
   <word>might've</word>
   <word>millennia</word>
   <word>mmm</word>
@@ -212,6 +215,7 @@
   <word>mojo</word>
   <word>molybdenum</word>
   <word>moonlet</word>
+  <word>multispectral</word>
   <word>must've</word>
   <word>nah</word>
   <word>namaste</word>
@@ -260,6 +264,8 @@
   <word>purée</word>
   <word>radium</word>
   <word>radon</word>
+  <word>reacclimate</word>
+  <word>reacclimating</word>
   <word>realise</word>
   <word>realised</word>
   <word>rebar</word>
@@ -382,11 +388,4 @@
   <word>yukina</word>
   <word>zinc</word>
   <word>zirconium</word>
-  <word>astrometry</word>
-  <word>esophaguses</word>
-  <word>micrografting</word>
-  <word>multispectral</word>
-  <word>reacclimate</word>
-  <word>reacclimating</word>
-  <word>skank</word>
 </words>

Dictionaries/eng_OCRFixReplaceList.xml

@@ -2953,7 +2953,6 @@
     <LinePart from=" hasn 't " to=" hasn't " />
     <LinePart from=" you ' re " to=" you're " />
     <LinePart from=" You ' re " to=" You're " />
-    <LinePart from=" that' s " to=" that's " />
     <LinePart from="1 2th " to="12th " />
     <LinePart from="-| " to="- I " />
     <LinePart from=" | " to=" I " />
@@ -3123,7 +3122,6 @@
     <Beginning from="She' s " to="She's " />
     <Beginning from="O kay, " to="Okay, " />
     <Beginning from="l didn't" to="I didn't" />
-    <Beginning from="l don't" to="I don't" />
     <Beginning from="-1 " to="- I" />
     <Beginning from="I'II " to="I'll " />
     <Beginning from="Do l " to="Do I " />

Dictionaries/hr_NoBreakAfterList.xml

@@ -1,28 +1,28 @@
-<NoBreakAfterList>
+<NoBreakAfterList>
   <Item>bl</Item>
   <Item>bl.</Item>
   <Item>dipl</Item>
   <Item>dipl.</Item>
-	<Item>Dr.</Item>
+  <Item>dr</Item>
   <Item>Dr</Item>
   <Item>dr.</Item>
-	<Item>dr</Item>
-	<Item>g.</Item>
+  <Item>Dr.</Item>
   <Item>g</Item>
-	<Item>god.</Item>
-	<Item>gosp</Item>
-	<Item>gosp.</Item>
+  <Item>g.</Item>
   <Item>gđa</Item>
-	<Item>gđu</Item>
   <Item>gđica</Item>
   <Item>gđice</Item>
   <Item>gđici</Item>
   <Item>gđicu</Item>
+  <Item>gđu</Item>
+  <Item>god.</Item>
+  <Item>gosp</Item>
+  <Item>gosp.</Item>
   <Item>ing.</Item>
   <Item>mr.</Item>
   <Item>prof</Item>
   <Item>prof.</Item>
-	<Item>sv.</Item>
   <Item>sv</Item>
+  <Item>sv.</Item>
   <Item>vlč.</Item>
 </NoBreakAfterList>

Dictionaries/hrv_OCRFixReplaceList.xml

@@ -682,9 +682,9 @@
     <Word from="ljetnje" to="ljetne" />
     <Word from="magnezijuma" to="magnezija" />
     <Word from="magnezijumu" to="magneziju" />
-    <Word from="maja" to="svibnja"/>
-    <Word from="maju" to="svibnju"/>
-    <Word from="majem" to="svibnjem"/>
+    <Word from="maja" to="svibnja" />
+    <Word from="maju" to="svibnju" />
+    <Word from="majem" to="svibnjem" />
     <Word from="majek" to="majke" />
     <Word from="majca" to="majica" />
     <Word from="majce" to="majice" />
@@ -695,9 +695,9 @@
     <Word from="maloprije" to="malo prije" />
     <Word from="manifestuje" to="manifestira" />
     <Word from="manifestuju" to="manifestiraju" />
-    <Word from="maram" to="moram"/>
+    <Word from="maram" to="moram" />
     <Word from="marta" to="ožujka" />
-    <Word from="martu" to="ožujku"/>
+    <Word from="martu" to="ožujku" />
     <Word from="martom" to="ožujkom" />
     <Word from="matori" to="stari" />
     <Word from="mehur" to="mjehur" />
@@ -1904,7 +1904,6 @@
     <Word from="Živeo" to="Živio" />
     <Word from="živeo" to="živio" />
     <Word from="žmureo" to="žmirio" />
-
     <!-- imena -->
     <Word from="Abi" to="Abby" />
     <Word from="Alis" to="Alice" />
@@ -3381,7 +3380,8 @@
     <RegEx find="obed" replaceWith="objed" />
     <RegEx find="obem" replaceWith="oblem" />
     <RegEx find="([oO])besi" replaceWith="$1bjesi" />
-    <RegEx find="([Oo])bs" replaceWith="$1ps" /><!-- Jobs je problem! -->
+    <!-- Jobs je problem! -->
+    <RegEx find="([Oo])bs" replaceWith="$1ps" />
     <RegEx find="o[cčć]aren" replaceWith="očaran" />
     <RegEx find="([oO])d([kp])" replaceWith="$1t$2" />
     <RegEx find="oćeju" replaceWith="oće" />
@@ -3525,7 +3525,6 @@
     <RegEx find="žalj?ev" replaceWith="žalijev" />
     <RegEx find="živać" replaceWith="živat ć" />
     <RegEx find="žive([lot])" replaceWith="živi$1" />
-
     <!-- osobna imena/prezimena i imena gradova/država itd. -->
     <RegEx find="Afghanistan" replaceWith="Afganistan" />
     <RegEx find="Ajdah" replaceWith="Idah" />

Dictionaries/spa_OCRFixReplaceList.xml

@@ -945,7 +945,6 @@
     <RegEx find="\Bo(log[ao]s?\b)" replaceWith="ó$1" />
     <RegEx find="\Ba(log[ao]s?\b)" replaceWith="á$1" />
     <RegEx find="\Bi(log[ao]s?\b)" replaceWith="í$1" />
-
     <RegEx find="\bIes\b" replaceWith="les" />
     <RegEx find="\bIos\b" replaceWith="los" />
   </RegularExpressions>

Dictionaries/sr_NoBreakAfterList.xml

@@ -1,28 +1,28 @@
-<NoBreakAfterList>
+<NoBreakAfterList>
   <Item>bl</Item>
   <Item>bl.</Item>
   <Item>dipl</Item>
   <Item>dipl.</Item>
-	<Item>Dr.</Item>
+  <Item>dr</Item>
   <Item>Dr</Item>
   <Item>dr.</Item>
-	<Item>dr</Item>
-	<Item>g.</Item>
+  <Item>Dr.</Item>
   <Item>g</Item>
-	<Item>god.</Item>
-	<Item>gosp</Item>
-	<Item>gosp.</Item>
+  <Item>g.</Item>
   <Item>gđa</Item>
-	<Item>gđu</Item>
   <Item>gđica</Item>
   <Item>gđice</Item>
   <Item>gđici</Item>
   <Item>gđicu</Item>
+  <Item>gđu</Item>
+  <Item>god.</Item>
+  <Item>gosp</Item>
+  <Item>gosp.</Item>
   <Item>ing.</Item>
   <Item>mr.</Item>
   <Item>prof</Item>
   <Item>prof.</Item>
-	<Item>sv.</Item>
   <Item>sv</Item>
+  <Item>sv.</Item>
   <Item>vlč.</Item>
 </NoBreakAfterList>