Updated xml dictionaries

git-svn-id: https://subtitleedit.googlecode.com/svn/trunk@787 99eadd0c-20b8-1223-b5c4-2a2b2df33de2
This commit is contained in:
niksedk 2011-11-06 11:08:22 +00:00
parent ca6d1b9cfc
commit a9a51bab03
5 changed files with 452 additions and 9 deletions

View File

@ -0,0 +1,56 @@
<words>
<word>geht's</word>
<word>ner</word>
<word>nerv</word>
<word>leid</word>
<word>einfühlsam</word>
<word>hab's</word>
<word>ok</word>
<word>krieg</word>
<word>vietnamesisch</word>
<word>wow</word>
<word>hm</word>
<word>mein's</word>
<word>wär's</word>
<word>sexspielzeug</word>
<word>rauszubringen</word>
<word>du's</word>
<word>vorbeigeschaut</word>
<word>zurückmachen</word>
<word>sorg</word>
<word>schwör</word>
<word>scheißegal</word>
<word>suhlen</word>
<word>scheißkahn</word>
<word>aah</word>
<word>15er</word>
<word>schmier</word>
<word>beeil</word>
<word>abschlafft</word>
<word>dauert's</word>
<word>verkackt</word>
<word>ho</word>
<word>ah</word>
<word>rausholen</word>
<word>ha</word>
<word>ähm</word>
<word>sich's</word>
<word>sie's</word>
<word>pack</word>
<word>draus</word>
<word>hmm</word>
<word>rumkriegen</word>
<word>verrücktesten</word>
<word>selbstverstümmelungen</word>
<word>80er</word>
<word>spuck</word>
<word>ich's</word>
<word>ähem</word>
<word>hey</word>
<word>na</word>
<word>deins</word>
<word>war's</word>
<word>ruh</word>
<word>tja</word>
<word>lehn</word>
</words>

View File

@ -0,0 +1,72 @@
<OCRFixReplaceList>
<WholeWords>
<Word from="suMM†" to="SUMMT" />
<Word from="Iehn" to="lehn" />
<Word from="ranzukommen" to="ran-zukommen" />
<Word from="gernhaben" to="gern-haben" />
<Word from="wiederhaben" to="wieder-haben" />
<Word from="fürchlerlichen" to="fürchterlichen" />
<Word from="runterspült" to="runter-spült" />
<Word from="Äu" to="Au" />
</WholeWords>
<PartialWordsAlways>
<!-- Will be replaced always -->
<WordPart from="¤" to="o" />
<WordPart from="IVI" to="M" />
<WordPart from="lVI" to="M" />
<WordPart from="IVl" to="M" />
<WordPart from="lVl" to="M" />
<WordPart from="IVl" to="M" />
<WordPart from="I\/I" to="M" />
<WordPart from="I\/I" to="M" />
<WordPart from="l\/I" to="M" />
<WordPart from="I\/l" to="M" />
<WordPart from="l\/l" to="M" />
<WordPart from="I\/l" to="M" />
</PartialWordsAlways>
<PartialWords>
<!-- Will be used to check words not in dictionary -->
<!-- If new word(s) exists in spelling dictionary, it(they) is accepted -->
<WordPart from="¤" to="o" />
<WordPart from="ı=" to="f" />
<WordPart from="fi" to="fi" />
<WordPart from="fl" to="fl" />
<WordPart from="/" to="l" />
<WordPart from="vv" to="w" />
<WordPart from="IVI" to="M" />
<WordPart from="lVI" to="M" />
<WordPart from="IVl" to="M" />
<WordPart from="lVl" to="M" />
<WordPart from="m" to="rn" />
<WordPart from="l" to="i" />
<WordPart from="€" to="e" />
<WordPart from="I" to="l" />
<WordPart from="c" to="o" />
<WordPart from="i" to="t" />
<WordPart from="cc" to="oo" />
<WordPart from="ii" to="tt" />
<WordPart from="ii" to="ü" />
<WordPart from="n/" to="ry" />
<!-- "f " will be two words -->
<WordPart from="f" to="f " />
<WordPart from="c" to="e" />
<WordPart from="o" to="e" />
<WordPart from="o" to="e" />
<WordPart from="I" to="t" />
<WordPart from="n" to="o" />
<WordPart from="s" to="e" />
<WordPart from="\A" to="Vi" />
<WordPart from="n/" to="rv" />
<WordPart from="í" to="i" />
<WordPart from="B" to="ß" />
</PartialWords>
<PartialLines>
<!-- Partial lines - but whole words only -->
<LinePart from="GEHT Auı=" to="GEHT AUF" />
</PartialLines>
<BeginLines>
</BeginLines>
<EndLines>
</EndLines>
<WholeLines />
</OCRFixReplaceList>

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>A.M.</name>
<name>Abolfazl</name>
@ -126,4 +126,94 @@
<name>Yousouf</name>
<name>Yutani</name>
<name>Zefram</name>
<name>1A</name>
<name>2A</name>
<name>Akashi</name>
<name>Atsuko</name>
<name>Bakuretsu</name>
<name>Benelli</name>
<name>Bifrost</name>
<name>Bleriot</name>
<name>Blyth</name>
<name>Botan</name>
<name>Branagh</name>
<name>Byakko</name>
<name>Chevallier</name>
<name>CONKLIN</name>
<name>D'Alessandro</name>
<name>Dennings</name>
<name>Foo Fighters</name>
<name>Furneaux</name>
<name>Genkai</name>
<name>Haris</name>
<name>Heimdall</name>
<name>Hiddleston</name>
<name>Iwamoto</name>
<name>Jo</name>
<name>Jotunheim</name>
<name>Kamiya</name>
<name>Kazuma</name>
<name>Keiko-chan</name>
<name>Kirishima</name>
<name>Kiyoshi</name>
<name>Koenma</name>
<name>Koenma-sama</name>
<name>Kreutz</name>
<name>Kuchinashi</name>
<name>Kurama</name>
<name>Kuwabara</name>
<name>Kuwabara-kun</name>
<name>Kuwabara-san</name>
<name>Kuwa-chan</name>
<name>Makaichu</name>
<name>Masaru</name>
<name>Merantau</name>
<name>Minamino</name>
<name>Mjolnir</name>
<name>Nadeshiko</name>
<name>Neuschwanstein</name>
<name>Nussbaum</name>
<name>Okubo</name>
<name>Pont Neuf</name>
<name>Puente Antiguo</name>
<name>Rei-gun</name>
<name>Reihado</name>
<name>Reikai</name>
<name>Reikai-ju</name>
<name>Riddick</name>
<name>Riom</name>
<name>Rubell</name>
<name>Sakashita</name>
<name>Sakyo</name>
<name>Sakyo-san</name>
<name>Sarayashiki</name>
<name>Sayaka</name>
<name>Setsuna</name>
<name>Sewell</name>
<name>Silverado</name>
<name>Skarsgård</name>
<name>Spectravision</name>
<name>Stellan</name>
<name>Stentz</name>
<name>Straczynski</name>
<name>Succinylcholine</name>
<name>Suzaku</name>
<name>Takenaka</name>
<name>Tarukane-san</name>
<name>Toguro</name>
<name>Urameshi</name>
<name>Urameshi-kun</name>
<name>Urameshi-san</name>
<name>VFX</name>
<name>Wazowski</name>
<name>Whedon</name>
<name>Yamada</name>
<name>Yanagisawa</name>
<name>Yokai</name>
<name>Yoki</name>
<name>Yukimura</name>
<name>Yukina-san</name>
<name>Yusuke</name>
<name>Zac</name>
<name>Zambarloukos</name>
</ignore_list>

View File

@ -1,4 +1,4 @@
<words>
<words>
<word>aeroplanes</word>
<word>ahh</word>
<word>ahhhh</word>
@ -122,4 +122,20 @@
<word>nothin</word>
<word>them's</word>
<word>mmm</word>
<word>giancarlo</word>
<word>gemensheidt</word>
<word>ajit</word>
<word>a.m.</word>
<word>p.m.</word>
<word>sci-fi</word>
<word>golconda</word>
<word>loupe</word>
<word>how're</word>
<word>mojo</word>
<word>9/11</word>
<word>sensei</word>
<word>hiei</word>
<word>kuwabara-kun</word>
<word>50s</word>
<word>yukina</word>
</words>

View File

@ -1,4 +1,4 @@
<OCRFixReplaceList>
<OCRFixReplaceList>
<WholeWords>
<Word from="/'ve" to="I've" />
<Word from="\/\/e're" to="We're" />
@ -997,6 +997,11 @@
<Word from="OId" to="Old" />
<Word from="CIark" to="Clark" />
<Word from="thatgoes" to="that goes" />
<Word from="llte" to="lite" />
<Word from="lnte" to="inte" />
<Word from="flra" to="fira" />
<Word from="vlll" to="vill" />
<Word from="dlg" to="dig" />
<Word from="BIood" to="Blood" />
<Word from="underyour" to="under your" />
<Word from="BIess" to="Bless" />
@ -1220,10 +1225,151 @@
<Word from="/'m" to="I'm" />
<Word from="Nlornin" to="Mornin" />
<Word from="novv" to="now" />
<Word from="Warcontinues" to="War continues" />
<Word from="bodiedyoung" to="bodied young" />
<Word from="ourbrave" to="our brave" />
<Word from="offreedom" to="of freedom" />
<Word from="toldyou" to="told you" />
<Word from="injust" to="in just" />
<Word from="shortyears" to="short years" />
<Word from="yourautomobile" to="your automobile" />
<Word from="dojust" to="do just" />
<Word from="Ourgoal" to="Our goal" />
<Word from="Hitlershare" to="Hitler share" />
<Word from="occultpowerand" to="occult power and" />
<Word from="superiorman" to="superior man" />
<Word from="hearme" to="hear me" />
<Word from="anotherstep" to="another step" />
<Word from="ofmicroinjections" to="of microinjections" />
<Word from="majormuscle" to="major muscle" />
<Word from="cellularchange" to="cellular change" />
<Word from="HerrSchmidt" to="Herr Schmidt" />
<Word from="Führerfeels" to="Führer feels" />
<Word from="forAmerica" to="for America" />
<Word from="orfall" to="or fall" />
<Word from="ofhis" to="of his" />
<Word from="oldAdolfon" to="old Adolf on" />
<Word from="thejaw" to="the jaw" />
<Word from="l'll" to="I'll" />
<Word from="Ml6" to="MI6" />
<Word from="isjust" to="is just" />
<Word from="Yourattention" to="Your attention" />
<Word from="furthernotice" to="further notice" />
<Word from="forNew" to="for New" />
<Word from="Don'tyou" to="Don't you" />
<Word from="L'd" to="I'd" />
<Word from="youngsterlike" to="youngster like" />
<Word from="theirfeet" to="their feet" />
<Word from="I/fe" to="life" />
<Word from="ea/'t/7" to="earth" />
<Word from="al/en" to="alien" />
<Word from="/mmerse" to="immerse" />
<Word from="Li/0/Academy" to="Lilly Academy" />
<Word from="gir/s'boarding" to="girls' boarding" />
<Word from="/ong" to="long" />
<Word from="rea/" to="real" />
<Word from="gir/" to="girl" />
<Word from="VVe" to="We" />
<Word from="Skarsgérd" to="Skarsgård" />
<Word from="VVinters" to="Winters" />
<Word from="couldn'tjust" to="couldn't just" />
<Word from="Iworked" to="I worked" />
<Word from="wnat's" to="what's" />
<Word from="tnatjust" to="that just" />
<Word from="getjust" to="get just" />
<Word from="Cruoially" to="Crucially" />
<Word from="can't_" to="Crucially" />
<Word from="d°9" to="dog" />
<Word from="/'/I" to="I'll" />
<Word from="We/I" to="Well" />
<Word from="yigg/mg" to="giggling" />
<Word from="Qpen" to="Open" />
<Word from="fingernail" to="fingernail" />
<Word from="inscfipfions" to="inscriptions" />
<Word from="0l/er" to="over" />
<Word from="shoufing" to="shouting" />
<Word from="Ithoguht" to="I thought" />
<Word from="okaY" to="okay" />
<Word from="yourhusband" to="your husband" />
<Word from="D0" to="Do" />
<Word from="lneed" to="l need" />
<Word from="DonWlook" to="Don't look" />
<Word from="Donllook" to="Don't look" />
<Word from="$COff$" to="scoffs" />
<Word from="z'housand" to="thousand" />
<Word from="shoufing" to="shouting" />
<Word from="thoguht" to="thought" />
<Word from="OKaY" to="Okay" />
<Word from="Véry" to="Very" />
<Word from="9/aim" to="alarm" />
<Word from="50rry" to="sorry" />
<Word from="dumbass" to="dumb-ass" />
<Word from="lwamoto" to="Iwamoto" />
<Word from="Ieapt" to="leapt" />
<Word from="yQu" to="you" />
<Word from="judgement" to="judgment" />
<Word from="Ithoughtl" to="I thought I" />
<Word from="Thatsand" to="That sand" />
<Word from="Yigg/mg" to="giggling" />
<Word from="$COff$" to="scoffs" />
<Word from="QkaY" to="Okay" />
<Word from="al/" to="all" />
<Word from="yummY" to="yummy" />
<Word from="Goddamit" to="Goddammit" />
<Word from="$COff$" to="scoffs" />
<Word from="inscfipfions" to="inscriptions" />
<Word from="9UnShQt" to="gunshot" />
<Word from="I/erilj/" to="verify" />
<Word from="P°P$" to="Pops" />
<Word from="I02" to="Pops" />
<Word from="pshyco" to="psycho" />
<Word from="SP/ash" to="Splash" />
<Word from="andleave" to="and leave" />
<Word from="Shoofing" to="Shooting" />
<Word from="VVhip" to="Whip" />
<Word from="kickflip" to="kick-flip" />
<Word from="wallplant" to="wall-plant" />
<Word from="NewJersey" to="New Jersey" />
<Word from="wasjust" to="was just" />
<Word from="furyou" to="for you" />
<Word from="Summerjust" to="Summer just" />
<Word from="if/just" to="if I just" />
<Word from="Isayt/1e" to="I say the" />
<Word from="We/came" to="Welcome" />
<Word from="he/ps" to="helps" />
<Word from="Persona/4/" to="Personally" />
<Word from="t/ying" to="trying" />
<Word from="p/ace" to="place" />
<Word from="surflaces" to="surfaces" />
<Word from="/ennifer" to="Jennifer" />
<Word from="comp/etey" to="completely" />
<Word from="jennifer" to="Jennifer" />
<Word from="wh/p" to="whip" />
<Word from="to//et" to="toilet" />
<Word from="buyyou" to="buy you" />
<Word from="P/ease" to="Please" />
<Word from="severa/parents" to="several parents" />
<Word from="wasjennifer" to="was Jennifer" />
<Word from="a/bum" to="album" />
<Word from="/t's" to="It's" />
<Word from="a//" to="all" />
<Word from="lworked" to="I worked" />
<Word from="thatjust" to="that just" />
<Word from="Thorjust" to="Thor just" />
<Word from="Exacfly" to="Exactly" />
<Word from="voicemail" to="voice-mail" />
<Word from="influence" to="influence" />
<Word from="Voilá" to="Voilà" />
<Word from="$'llOp" to="Stop" />
<Word from="Wh°ops" to="Whoops" />
<Word from="QUYS" to="GUYS" />
<Word from="I'E'$" to="It's" />
<Word from="YELL$" to="YELLS" />
<Word from="SIGH$" to="SIGHS" />
<Word from="S'EOp" to="Stop" />
</WholeWords>
<PartialWordsAlways>
<!-- Will be used to check words not in dictionary -->
<!-- If new word(s) exists in spelling dictionary, it(they) is accepted -->
<!-- Will be replaced always -->
<WordPart from="¤" to="o" />
<WordPart from="IVI" to="M" />
<WordPart from="lVI" to="M" />
@ -1238,8 +1384,8 @@
<WordPart from="I\/l" to="M" />
</PartialWordsAlways>
<PartialWords>
<!-- Will be used to check words not in dictionary -->
<!-- If new word(s) exists in spelling dictionary, it(they) is accepted -->
<!-- Will be used to check words not in dictionary.
If new word(s) and longer than 5 chars and exists in spelling dictionary, it is (or they are) accepted -->
<WordPart from="/" to="l" />
<WordPart from="vv" to="w" />
<WordPart from="m" to="rn" />
@ -1261,8 +1407,12 @@
<WordPart from="s" to="e" />
<WordPart from="l-l" to="H" />
<WordPart from="l_" to="L" />
<WordPart from="°" to="o" />
<WordPart from=")/" to="y" />
<WordPart from=")'" to="y" />
</PartialWords>
<PartialLines>
<!-- Partial lines - but whole words only -->
<LinePart from="at Hrst" to="at first" />
<LinePart from="can 't" to="can't" />
<LinePart from="come 0n" to="come on" />
@ -1329,9 +1479,57 @@
<LinePart from="Wal-I\/Iart" to="Wal-Mart" />
<LinePart from=" I']I " to=" I'll " />
<LinePart from="REPEA TING " to="REPEATING " />
<LinePart from="Mr_ " to="Mr. " />
<LinePart from="Let me 90" to="Let me go" />
<LinePart from="let me 90" to="let me go" />
<LinePart from="Let's 90" to="Let's go" />
<LinePart from="let's 90" to="let's go" />
<LinePart from="K)/le" to="Kyle" />
<LinePart from="you QUYS" to="you guys" />
<LinePart from="I 'rn " to="I'm " />
<LinePart from="N02' " to="Not " />
<LinePart from="I02'" to="lot" />
<LinePart from="doesn '1' " to="doesn't " />
<LinePart from="doesn'1' " to="doesn't " />
<LinePart from="don '1' " to="don't " />
<LinePart from="don'1' " to="don't " />
<LinePart from="Wasn '1' " to="Wasn't " />
<LinePart from="wasn '1' " to="wasn't " />
<LinePart from="doesn '1 " to="doesn't " />
<LinePart from="doesn'1 " to="doesn't " />
<LinePart from="don '1 " to="don't " />
<LinePart from="don'1 " to="don't " />
<LinePart from="Wasn '1 " to="Wasn't " />
<LinePart from="wasn '1 " to="wasn't " />
<LinePart from="B ullshiz'" to="Bullshit" />
<LinePart from="wem' off" to="went off" />
<LinePart from="/ need" to="I need" />
<LinePart from="I0 ve her" to="love her" />
<LinePart from="I0 ve you" to="love you" />
<LinePart from="big lunk" to="love you" />
<LinePart from="wasn 't" to="wasn't" />
<LinePart from="wouldn 't" to="wouldn't" />
<LinePart from="Wasn 't" to="Wasn't" />
<LinePart from="Wouldn 't" to="Wouldn't" />
<LinePart from="He)/" to="Hey" />
<LinePart from="he)/" to="hey" />
<LinePart from="Sa y" to="Say" />
<LinePart from="L ook" to="Look" />
<LinePart from="-ls it" to="- Is it" />
<LinePart from="-ls that" to="- Is that" />
<LinePart from="-ls this" to="- Is this" />
<LinePart from="-ls she" to="- Is she" />
<LinePart from="-ls he" to="- Is he" />
<LinePart from="ls it" to="Is it" />
<LinePart from="ls that" to="Is that" />
<LinePart from="ls this" to="Is this" />
<LinePart from="ls she" to="Is she" />
<LinePart from="ls he" to="Is he" />
<LinePart from="ls anyone" to="Is anyone" />
<LinePart from="y()u" to="you" />
<LinePart from="CHA TTERING" to="CHATTERING" />
</PartialLines>
<BeginLines>
<Beginning from="No w, " to="Now, " />
<Beginning from="l... I " to="I... I " />
<Beginning from="L... I " to="I... I " />
<Beginning from="lrn gonna" to="I'm gonna" />
@ -1386,15 +1584,26 @@
<Beginning from="L' m " to="I'm " />
<Beginning from="Lt' s " to="It's " />
<Beginning from="I']I " to="I'll " />
<Beginning from="...ls " to="...Is " />
<Beginning from="- ls " to="- Is " />
<Beginning from="...l " to="...I " />
<Beginning from="Ill " to="I'll " />
<Beginning from="L hope " to="I hope " />
<Beginning from="|E'$ " to="It's " />
</BeginLines>
<EndLines>
<Ending from=", sin" to=", sir." />
<Ending from=" mothen" to="mother." />
<Ending from=" mothen" to=" mother." />
<Ending from=" can't_" to=" can't." />
<Ending from=" openiL" to=" open it." />
<Ending from=" offl" to=" off!" />
<Ending from="pshycol" to="pshyco!" />
</EndLines>
<WholeLines>
<!-- Whole lines - including -" etc -->
<Line from="N0." to="No." />
<Line from="N0." to="No." />
<Line from="-N0." to="-No." />
<Line from="Noll" to="No!!" />
</WholeLines>
</OCRFixReplaceList>