Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-10-26 23:12:38 +02:00
Code Issues Projects Releases Wiki Activity

dictionaries: automated XML upkeep

Browse Source
This commit is contained in:
Waldi Ravens 2015-03-30 12:55:17 +02:00
parent 71d9d1476b
commit 10ecc81c67
18 changed files with 225 additions and 233 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Dictionaries/da_DK_user.xml
View File

@ -1,4 +1,4 @@
<words>
<words>
<word>bordtennis</word>
<word>dødtravlt</word>
<word>dérinde</word>

2
Dictionaries/da_NoBreakAfterList.xml
View File

@ -1,4 +1,4 @@
<NoBreakAfterList>
<NoBreakAfterList>
<Item> at</Item>
<Item> en</Item>
<Item> er</Item>

8
Dictionaries/dan_OCRFixReplaceList.xml
View File

@ -1,4 +1,4 @@
<OCRFixReplaceList>
<OCRFixReplaceList>
<WholeWords>
<Word from="Haner" to="Han er" />
<Word from="JaveL" to="Javel" />
@ -623,9 +623,7 @@
<WordPart from="í" to="i" />
</PartialWords>
<PartialLines />
<BeginLines>
</BeginLines>
<EndLines>
</EndLines>
<BeginLines />
<EndLines />
<WholeLines />
</OCRFixReplaceList>

9
Dictionaries/deu_OCRFixReplaceList.xml
View File

@ -6858,10 +6858,7 @@
<PartialLines>
<!-- Partial lines - but whole words only -->
</PartialLines>
<BeginLines>
</BeginLines>
<EndLines>
</EndLines>
<WholeLines>
</WholeLines>
<BeginLines />
<EndLines />
<WholeLines />
</OCRFixReplaceList>

2
Dictionaries/en_US_names_etc.xml
View File

@ -1,4 +1,4 @@
<!-- This list contains names/words with specific casing - and specific to English only -->
<!-- This list contains names/words with specific casing - and specific to English only -->
<ignore_list>
<name>Abbey</name>
<name>Abbie</name>

2
Dictionaries/en_US_user.xml
View File

@ -1,4 +1,4 @@
<words>
<words>
<word>50s</word>
<word>9/11</word>
<word>a.m.</word>

4
Dictionaries/es_MX_user.xml
View File

@ -1,4 +1,4 @@
<words>
<words>
<word>recordarle</word>
<word>montos</word>
<word>estresados</word>
@ -831,4 +831,4 @@
<word>estafarme</word>
<word>saloncito</word>
<word>imagínatelo</word>
</words>
</words>

4
Dictionaries/es_NoBreakAfterList.xml
View File

@ -1,4 +1,4 @@
<NoBreakAfterList>
<NoBreakAfterList>
<Item RegEx="True"> [¿¡«]</Item>
<Item RegEx="True"> [A-ZÁ-ÚÑ]</Item>
<Item> a</Item>
@ -112,4 +112,4 @@
<Item> en la</Item>
<Item> cuán</Item>
<Item> más</Item>
</NoBreakAfterList>
</NoBreakAfterList>

2
Dictionaries/fin_OCRFixReplaceList.xml
View File

@ -1,4 +1,4 @@
<ReplaceList>
<ReplaceList>
<WholeWords>
<Word from="kellojo" to="kello jo" />
<Word from="onjo" to="on jo" />

2
Dictionaries/fra_OCRFixReplaceList.xml
View File

@ -1,4 +1,4 @@
<ReplaceList>
<ReplaceList>
<WholeWords>
<Word from="@immatriculation" to="d'immatriculation" />
<Word from="acquer" to="acquér" />

396
Dictionaries/hrv_OCRFixReplaceList.xml
View File

@ -1,4 +1,4 @@
<ReplaceList>
<ReplaceList>
<WholeWords>
<Word from="čovek" to="čovjek" />
<Word from="podnjeti" to="podnijeti" />
@ -39,179 +39,179 @@
<Word from="vrednost" to="vrijednost" />
<Word from="funkcioniše" to="funkcionira" />
<Word from="skuvati" to="skuhati" />
<Word from="napolje" to="van" />
<Word from="deli" to="dijeli" />
<Word from="pesmu" to="pjesmu" />
<Word from="ovde" to="ovdje" />
<Word from="pre" to="prije" />
<Word from="obe" to="obje" />
<Word from="reči" to="riječi" />
<Word from="reč" to="riječ" />
<Word from="nauka" to="znanost" />
<Word from="smem" to="smijem" />
<Word from="gde" to="gdje" />
<Word from="smao" to="samo" />
<Word from="šta" to="što" />
<Word from="štp" to="što" />
<Word from="niko" to="nitko" />
<Word from="koa" to="kao" />
<Word from="haos" to="kaos" />
<Word from="ker" to="pas" />
<Word from="kera" to="psa" />
<Word from="keru" to="psu" />
<Word from="levi" to="lijevi" />
<Word from="leva" to="lijeva" />
<Word from="preterao" to="pretjerao" />
<Word from="oceni" to="ocijeni" />
<Word from="izleteo" to="izletio" />
<Word from="razumeš" to="razumiješ" />
<Word from="razumeo" to="razumio" />
<Word from="osećaj" to="osjećaj" />
<Word from="večan" to="vječan" />
<Word from="srećan" to="sretan" />
<Word from="srećna" to="sretna" />
<Word from="srećno" to="sretno" />
<Word from="promena" to="promjena" />
<Word from="pogrešio" to="pogriješio" />
<Word from="pogrješio" to="pogriješio" />
<Word from="htela" to="htjela" />
<Word from="hteo" to="htio" />
<Word from="uvek" to="uvijek" />
<Word from="pantalone" to="hlače" />
<Word from="umetnost" to="umjetnost" />
<Word from="umetnik" to="umjetnik" />
<Word from="umetnički" to="umjetnički" />
<Word from="smešten" to="smješten" />
<Word from="reko" to="rekao" />
<Word from="veoma" to="vrlo" />
<Word from="nauka" to="znanost" />
<Word from="kolena" to="koljena" />
<Word from="voleo" to="volio" />
<Word from="volena" to="voljena" />
<Word from="uvek" to="uvijek" />
<Word from="delo" to="djelo" />
<Word from="dole" to="dolje" />
<Word from="dešava" to="događa" />
<Word from="prihvatam" to="prihvaćam" />
<Word from="vratiće" to="vratit će" />
<Word from="đavola" to="vraga" />
<Word from="ćale" to="tata" />
<Word from="uspe" to="uspije" />
<Word from="uspelo" to="uspjelo" />
<Word from="plata" to="plaća" />
<Word from="raspust" to="odmor" />
<Word from="ujutru" to="ujutro" />
<Word from="šagarepa" to="mrkva" />
<Word from="opšta" to="opća" />
<Word from="paramparčad" to="komadići" />
<Word from="uopšte" to="uopće" />
<Word from="nervi" to="živci" />
<Word from="lek" to="lijek" />
<Word from="reka" to="rijeka" />
<Word from="zamena" to="zamjena" />
<Word from="zameniti" to="zamijeniti" />
<Word from="zamjeniti" to="zamijeniti" />
<Word from="promeniti" to="promijeniti" />
<Word from="nedelja" to="nedjelja" />
<Word from="ispričaći" to="ispričati ću" />
<Word from="misliće" to="mislit će" />
<Word from="obezbediti" to="osigurati" />
<Word from="obezbeđenje" to="osiguranje" />
<Word from="obezbeđenju" to="osiguranju" />
<Word from="pomerati" to="micati" />
<Word from="pomera" to="miče" />
<Word from="pomeraj" to="miči" />
<Word from="bezbjedna" to="sigurna" />
<Word from="dete" to="dijete" />
<Word from="djete" to="dijete" />
<Word from="svetlost" to="svjetlost" />
<Word from="svetlo" to="svjetlo" />
<Word from="žlezda" to="žlijezda" />
<Word from="žljezda" to="žlijezda" />
<Word from="medved" to="medvjed" />
<Word from="merač" to="mjerač" />
<Word from="sočivo" to="leća" />
<Word from="sočiva" to="leće" />
<Word from="komandni" to="zapovjedni" />
<Word from="sećam" to="sjećam" />
<Word from="sirće" to="ocat" />
<Word from="sirćetu" to="octu" />
<Word from="vetar" to="vjetar" />
<Word from="vjetru" to="vetru" />
<Word from="vetra" to="vjetra" />
<Word from="učestvuje" to="sudjeluje" />
<Word from="prosečno" to="prosječno" />
<Word from="tačka" to="točka" />
<Word from="tačke" to="točke" />
<Word from="tačno" to="točno" />
<Word from="tačna" to="točna" />
<Word from="telo" to="tijelo" />
<Word from="dve" to="dvije" />
<Word from="deo" to="dio" />
<Word from="delu" to="dijelu" />
<Word from="deluje" to="djeluje" />
<Word from="uglom" to="kutom" />
<Word from="verovatno" to="vjerojatno" />
<Word from="neverovatno" to="nevjerojatno" />
<Word from="talas" to="val" />
<Word from="napraviću" to="napravit ću" />
<Word from="bićete" to="bit ćete" />
<Word from="kompanija" to="tvrtka" />
<Word from="testera" to="pila" />
<Word from="sneg" to="snijeg" />
<Word from="smeš" to="smiješ" />
<Word from="smešak" to="smješak" />
<Word from="stepen" to="stupanj" />
<Word from="stepeni" to="stupnjeva" />
<Word from="smešno" to="smiješno" />
<Word from="bežanje" to="bježanje" />
<Word from="prihvatanje" to="prihvaćanje" />
<Word from="obeležavanje" to="obilježavanje" />
<Word from="spoljašnji" to="vanjski" />
<Word from="kontrolisano" to="kontrolirano" />
<Word from="nekontrolisano" to="nekontrolirano" />
<Word from="džanki" to="ovisnik" />
<Word from="prevod" to="prijevod" />
<Word from="umesto" to="umjesto" />
<Word from="ignoriši" to="ignoriraj" />
<Word from="kučka" to="kuja" />
<Word from="ubica" to="ubojica" />
<Word from="lično" to="osobno" />
<Word from="lična" to="osobna" />
<Word from="dobo" to="dobro" />
<Word from="meša" to="miješa" />
<Word from="ceo" to="cijeli" />
<Word from="dečak" to="dječak" />
<Word from="foka" to="tuljan" />
<Word from="foke" to="tuljani" />
<Word from="kafu" to="kavu" />
<Word from="kafa" to="kava" />
<Word from="kupatilo" to="kupaona" />
<Word from="kupatilu" to="kupaoni" />
<Word from="historija" to="povijest" />
<Word from="historiju" to="povijest" />
<Word from="historije" to="povijesti" />
<Word from="precednik" to="predsjednik" />
<Word from="precjednik" to="predsjednik" />
<Word from="precijednik" to="predsjednik" />
<Word from="uspe" to="uspije" />
<Word from="menja" to="mijenja" />
<Word from="uspeju" to="uspiju" />
<Word from="poenta" to="poanta" />
<Word from="poente" to="poante" />
<Word from="štampa" to="tisak" />
<Word from="džigerica" to="jetra" />
<Word from="džigericu" to="jetru" />
<Word from="sistem" to="sustav" />
<Word from="sistemu" to="sustavu" />
<Word from="sistemi" to="sustavi" />
<Word from="udavit" to="utopit" />
<Word from="dalmatinac" to="Dalmatinac" />
<Word from="zagorac" to="Zagorac" />
<Word from="prigorac" to="Prigorac" />
<Word from="slavonac" to="Slavonac" />
<Word from="Pol" to="Paul" />
<Word from="Ejmi" to="Aimee" />
<Word from="napolje" to="van" />
<Word from="deli" to="dijeli" />
<Word from="pesmu" to="pjesmu" />
<Word from="ovde" to="ovdje" />
<Word from="pre" to="prije" />
<Word from="obe" to="obje" />
<Word from="reči" to="riječi" />
<Word from="reč" to="riječ" />
<Word from="nauka" to="znanost" />
<Word from="smem" to="smijem" />
<Word from="gde" to="gdje" />
<Word from="smao" to="samo" />
<Word from="šta" to="što" />
<Word from="štp" to="što" />
<Word from="niko" to="nitko" />
<Word from="koa" to="kao" />
<Word from="haos" to="kaos" />
<Word from="ker" to="pas" />
<Word from="kera" to="psa" />
<Word from="keru" to="psu" />
<Word from="levi" to="lijevi" />
<Word from="leva" to="lijeva" />
<Word from="preterao" to="pretjerao" />
<Word from="oceni" to="ocijeni" />
<Word from="izleteo" to="izletio" />
<Word from="razumeš" to="razumiješ" />
<Word from="razumeo" to="razumio" />
<Word from="osećaj" to="osjećaj" />
<Word from="večan" to="vječan" />
<Word from="srećan" to="sretan" />
<Word from="srećna" to="sretna" />
<Word from="srećno" to="sretno" />
<Word from="promena" to="promjena" />
<Word from="pogrešio" to="pogriješio" />
<Word from="pogrješio" to="pogriješio" />
<Word from="htela" to="htjela" />
<Word from="hteo" to="htio" />
<Word from="uvek" to="uvijek" />
<Word from="pantalone" to="hlače" />
<Word from="umetnost" to="umjetnost" />
<Word from="umetnik" to="umjetnik" />
<Word from="umetnički" to="umjetnički" />
<Word from="smešten" to="smješten" />
<Word from="reko" to="rekao" />
<Word from="veoma" to="vrlo" />
<Word from="nauka" to="znanost" />
<Word from="kolena" to="koljena" />
<Word from="voleo" to="volio" />
<Word from="volena" to="voljena" />
<Word from="uvek" to="uvijek" />
<Word from="delo" to="djelo" />
<Word from="dole" to="dolje" />
<Word from="dešava" to="događa" />
<Word from="prihvatam" to="prihvaćam" />
<Word from="vratiće" to="vratit će" />
<Word from="đavola" to="vraga" />
<Word from="ćale" to="tata" />
<Word from="uspe" to="uspije" />
<Word from="uspelo" to="uspjelo" />
<Word from="plata" to="plaća" />
<Word from="raspust" to="odmor" />
<Word from="ujutru" to="ujutro" />
<Word from="šagarepa" to="mrkva" />
<Word from="opšta" to="opća" />
<Word from="paramparčad" to="komadići" />
<Word from="uopšte" to="uopće" />
<Word from="nervi" to="živci" />
<Word from="lek" to="lijek" />
<Word from="reka" to="rijeka" />
<Word from="zamena" to="zamjena" />
<Word from="zameniti" to="zamijeniti" />
<Word from="zamjeniti" to="zamijeniti" />
<Word from="promeniti" to="promijeniti" />
<Word from="nedelja" to="nedjelja" />
<Word from="ispričaći" to="ispričati ću" />
<Word from="misliće" to="mislit će" />
<Word from="obezbediti" to="osigurati" />
<Word from="obezbeđenje" to="osiguranje" />
<Word from="obezbeđenju" to="osiguranju" />
<Word from="pomerati" to="micati" />
<Word from="pomera" to="miče" />
<Word from="pomeraj" to="miči" />
<Word from="bezbjedna" to="sigurna" />
<Word from="dete" to="dijete" />
<Word from="djete" to="dijete" />
<Word from="svetlost" to="svjetlost" />
<Word from="svetlo" to="svjetlo" />
<Word from="žlezda" to="žlijezda" />
<Word from="žljezda" to="žlijezda" />
<Word from="medved" to="medvjed" />
<Word from="merač" to="mjerač" />
<Word from="sočivo" to="leća" />
<Word from="sočiva" to="leće" />
<Word from="komandni" to="zapovjedni" />
<Word from="sećam" to="sjećam" />
<Word from="sirće" to="ocat" />
<Word from="sirćetu" to="octu" />
<Word from="vetar" to="vjetar" />
<Word from="vjetru" to="vetru" />
<Word from="vetra" to="vjetra" />
<Word from="učestvuje" to="sudjeluje" />
<Word from="prosečno" to="prosječno" />
<Word from="tačka" to="točka" />
<Word from="tačke" to="točke" />
<Word from="tačno" to="točno" />
<Word from="tačna" to="točna" />
<Word from="telo" to="tijelo" />
<Word from="dve" to="dvije" />
<Word from="deo" to="dio" />
<Word from="delu" to="dijelu" />
<Word from="deluje" to="djeluje" />
<Word from="uglom" to="kutom" />
<Word from="verovatno" to="vjerojatno" />
<Word from="neverovatno" to="nevjerojatno" />
<Word from="talas" to="val" />
<Word from="napraviću" to="napravit ću" />
<Word from="bićete" to="bit ćete" />
<Word from="kompanija" to="tvrtka" />
<Word from="testera" to="pila" />
<Word from="sneg" to="snijeg" />
<Word from="smeš" to="smiješ" />
<Word from="smešak" to="smješak" />
<Word from="stepen" to="stupanj" />
<Word from="stepeni" to="stupnjeva" />
<Word from="smešno" to="smiješno" />
<Word from="bežanje" to="bježanje" />
<Word from="prihvatanje" to="prihvaćanje" />
<Word from="obeležavanje" to="obilježavanje" />
<Word from="spoljašnji" to="vanjski" />
<Word from="kontrolisano" to="kontrolirano" />
<Word from="nekontrolisano" to="nekontrolirano" />
<Word from="džanki" to="ovisnik" />
<Word from="prevod" to="prijevod" />
<Word from="umesto" to="umjesto" />
<Word from="ignoriši" to="ignoriraj" />
<Word from="kučka" to="kuja" />
<Word from="ubica" to="ubojica" />
<Word from="lično" to="osobno" />
<Word from="lična" to="osobna" />
<Word from="dobo" to="dobro" />
<Word from="meša" to="miješa" />
<Word from="ceo" to="cijeli" />
<Word from="dečak" to="dječak" />
<Word from="foka" to="tuljan" />
<Word from="foke" to="tuljani" />
<Word from="kafu" to="kavu" />
<Word from="kafa" to="kava" />
<Word from="kupatilo" to="kupaona" />
<Word from="kupatilu" to="kupaoni" />
<Word from="historija" to="povijest" />
<Word from="historiju" to="povijest" />
<Word from="historije" to="povijesti" />
<Word from="precednik" to="predsjednik" />
<Word from="precjednik" to="predsjednik" />
<Word from="precijednik" to="predsjednik" />
<Word from="uspe" to="uspije" />
<Word from="menja" to="mijenja" />
<Word from="uspeju" to="uspiju" />
<Word from="poenta" to="poanta" />
<Word from="poente" to="poante" />
<Word from="štampa" to="tisak" />
<Word from="džigerica" to="jetra" />
<Word from="džigericu" to="jetru" />
<Word from="sistem" to="sustav" />
<Word from="sistemu" to="sustavu" />
<Word from="sistemi" to="sustavi" />
<Word from="udavit" to="utopit" />
<Word from="dalmatinac" to="Dalmatinac" />
<Word from="zagorac" to="Zagorac" />
<Word from="prigorac" to="Prigorac" />
<Word from="slavonac" to="Slavonac" />
<Word from="Pol" to="Paul" />
<Word from="Ejmi" to="Aimee" />
</WholeWords>
<PartialWords>
<WordPart from="devojk" to="djevojk" />
@ -225,32 +225,32 @@
<WordPart from="promen" to="promijen" />
<WordPart from="posledic" to="posljedic" />
<WordPart from="poslednj" to="posljednj" />
<WordPart from="zvaničn" to="služben" />
<WordPart from="meseč" to="mjeseč" />
<WordPart from="grejanj" to="grijanj" />
<WordPart from="sačeka" to="pričeka" />
<WordPart from="kelner" to="konobar" />
<WordPart from="evrop" to="europ" />
<WordPart from="pacov" to="štakor" />
<WordPart from="historijsk" to="povijesn" />
<WordPart from="izbegava" to="izbjegava" />
<WordPart from="uspeh" to="uspjeh" />
<WordPart from="uslov" to="uvjet" />
<WordPart from="vazduh" to="zrak" />
<WordPart from="oseća" to="osjeća" />
<WordPart from="svež" to="svjež" />
<WordPart from="smešn" to="smiješn" />
<WordPart from="peškir" to="ručnik" />
<WordPart from="odeć" to="odjeć" />
<WordPart from="pozorišt" to="kazališt" />
<WordPart from="naučn" to="znanstven" />
<WordPart from="vredi" to="vrijedi" />
<WordPart from="zvaničn" to="služben" />
<WordPart from="meseč" to="mjeseč" />
<WordPart from="grejanj" to="grijanj" />
<WordPart from="sačeka" to="pričeka" />
<WordPart from="kelner" to="konobar" />
<WordPart from="evrop" to="europ" />
<WordPart from="pacov" to="štakor" />
<WordPart from="historijsk" to="povijesn" />
<WordPart from="izbegava" to="izbjegava" />
<WordPart from="uspeh" to="uspjeh" />
<WordPart from="uslov" to="uvjet" />
<WordPart from="vazduh" to="zrak" />
<WordPart from="oseća" to="osjeća" />
<WordPart from="svež" to="svjež" />
<WordPart from="smešn" to="smiješn" />
<WordPart from="peškir" to="ručnik" />
<WordPart from="odeć" to="odjeć" />
<WordPart from="pozorišt" to="kazališt" />
<WordPart from="naučn" to="znanstven" />
<WordPart from="vredi" to="vrijedi" />
</PartialWords>
<BeginLines />
<EndLines />
<WholeLines>
<Line from="Dobro veče" to="Dobra večer" />
<Line from="Dobro večer" to="Dobra večer" />
<Line from="Dobar večer" to="Dobra večer" />
<Line from="Dobro veče" to="Dobra večer" />
<Line from="Dobro večer" to="Dobra večer" />
<Line from="Dobar večer" to="Dobra večer" />
</WholeLines>
</ReplaceList>
</ReplaceList>

9
Dictionaries/nor_OCRFixReplaceList.xml
View File

@ -1,6 +1,5 @@
<OCRFixReplaceList>
<WholeWords>
</WholeWords>
<WholeWords />
<PartialWords>
<!-- Will be used to check words not in dictionary -->
<!-- If new word(s) exists in spelling dictionary, it(they) is accepted -->
@ -35,9 +34,7 @@
<WordPart from="í" to="i" />
</PartialWords>
<PartialLines />
<BeginLines>
</BeginLines>
<EndLines>
</EndLines>
<BeginLines />
<EndLines />
<WholeLines />
</OCRFixReplaceList>

4
Dictionaries/por_OCRFixReplaceList.xml
View File

@ -1,4 +1,4 @@
<OCRFixReplaceList>
<OCRFixReplaceList>
<WholeWords>
<Word from="abitual" to="habitual" />
<Word from="àcerca" to="acerca" />
@ -522,4 +522,4 @@
<RegEx find="\Bdmnis\B" replaceWith="dminis" />
<RegEx find="\Blcól\B" replaceWith="lcoól" />
</RegularExpressions>
</OCRFixReplaceList>
</OCRFixReplaceList>

2
Dictionaries/pt_PT_user.xml
View File

@ -1,4 +1,4 @@
<words>
<words>
<word>o</word>
<word>é</word>
<word>e</word>

2
Dictionaries/ru_RU_user.xml
View File

@ -1,4 +1,4 @@
<words>
<words>
<word>А</word>
<word>Ага</word>
<word>Ай</word>

2
Dictionaries/rus_OCRFixReplaceList.xml
View File

@ -1,4 +1,4 @@
<ReplaceList>
<ReplaceList>
<WholeWords>
<Word from="НЄЙ" to="НЕЙ" />
<Word from="ОРГЗНИЗМОБ" to="ОРГАНИЗМА" />

4
Dictionaries/spa_OCRFixReplaceList.xml
View File

@ -1,4 +1,4 @@
<ReplaceList>
<ReplaceList>
<WholeWords>
<!-- Abreviaturas simples -->
<Word from="KBs" to="kB" />
@ -955,4 +955,4 @@
<RegEx find="\Ba(log[ao]s?\b)" replaceWith="á$1" />
<RegEx find="\Bi(log[ao]s?\b)" replaceWith="í$1" />
</RegularExpressions>
</ReplaceList>
</ReplaceList>

2
Dictionaries/swe_OCRFixReplaceList.xml
View File

@ -1,4 +1,4 @@
<ReplaceList>
<ReplaceList>
<WholeWords>
<Word from="lârt" to="lärt" />
<Word from="hedervårda" to="hedervärda" />