larvitgeodata
Version:
Geo data, primarily ISO territories, languages etc. Data fetched mostly from CLDR.
391 lines (389 loc) • 14.8 kB
text/xml
<!--
Copyright © 1991-2013 Unicode, Inc.
CLDR data files are interpreted according to the LDML specification (http://unicode.org/reports/tr35/)
For terms of use, see http://www.unicode.org/copyright.html
-->
<supplementalData>
<version number="$Revision: 11914 $"/>
<transforms>
<transform source="Latin" target="InterIndic" direction="forward" visibility="internal">
<comment># Latin-InterIndic</comment>
<comment>#:: NFD;</comment>
<comment>#\u0E00 reserved</comment>
<comment>#consonants</comment>
<tRule>$chandrabindu=\uE001;</tRule>
<tRule>$anusvara=\uE002;</tRule>
<tRule>$visarga=\uE003;</tRule>
<comment>#\u0E004 reserved</comment>
<comment># w←vowel→ represents the stand-alone form</comment>
<tRule>$wa=\uE005;</tRule>
<tRule>$waa=\uE006;</tRule>
<tRule>$wi=\uE007;</tRule>
<tRule>$wii=\uE008;</tRule>
<tRule>$wu=\uE009;</tRule>
<tRule>$wuu=\uE00A;</tRule>
<tRule>$wr=\uE00B;</tRule>
<tRule>$wl=\uE00C;</tRule>
<tRule>$wce=\uE00D; # LETTER CANDRA E</tRule>
<tRule>$wse=\uE00E; # LETTER SHORT E</tRule>
<tRule>$we=\uE00F; # ए LETTER E</tRule>
<tRule>$wai=\uE010;</tRule>
<tRule>$wco=\uE011; # LETTER CANDRA O</tRule>
<tRule>$wso=\uE012; # LETTER SHORT O</tRule>
<tRule>$wo=\uE013; # ओ LETTER O</tRule>
<tRule>$wau=\uE014;</tRule>
<tRule>$ka=\uE015;</tRule>
<tRule>$kha=\uE016;</tRule>
<tRule>$ga=\uE017;</tRule>
<tRule>$gha=\uE018;</tRule>
<tRule>$nga=\uE019;</tRule>
<tRule>$ca=\uE01A;</tRule>
<tRule>$cha=\uE01B;</tRule>
<tRule>$ja=\uE01C;</tRule>
<tRule>$jha=\uE01D;</tRule>
<tRule>$nya=\uE01E;</tRule>
<tRule>$tta=\uE01F;</tRule>
<tRule>$ttha=\uE020;</tRule>
<tRule>$dda=\uE021;</tRule>
<tRule>$ddha=\uE022;</tRule>
<tRule>$nna=\uE023;</tRule>
<tRule>$ta=\uE024;</tRule>
<tRule>$tha=\uE025;</tRule>
<tRule>$da=\uE026;</tRule>
<tRule>$dha=\uE027;</tRule>
<tRule>$na=\uE028;</tRule>
<tRule>$ena=\uE029; #compatibility</tRule>
<tRule>$pa=\uE02A;</tRule>
<tRule>$pha=\uE02B;</tRule>
<tRule>$ba=\uE02C;</tRule>
<tRule>$bha=\uE02D;</tRule>
<tRule>$ma=\uE02E;</tRule>
<tRule>$ya=\uE02F;</tRule>
<tRule>$ra=\uE030;</tRule>
<tRule>$rra=\uE031;</tRule>
<tRule>$la=\uE032;</tRule>
<tRule>$lla=\uE033;</tRule>
<tRule>$ela=\uE034; #compatibility</tRule>
<tRule>$va=\uE035;</tRule>
<tRule>$vva=\uE081;</tRule>
<tRule>$sha=\uE036;</tRule>
<tRule>$ssa=\uE037;</tRule>
<tRule>$sa=\uE038;</tRule>
<tRule>$ha=\uE039;</tRule>
<comment>#\u093A Reserved</comment>
<comment>#\u093B Reserved</comment>
<tRule>$nukta=\uE03C;</tRule>
<tRule>$avagraha=\uE03D; # SIGN AVAGRAHA</tRule>
<comment># ←vowel→ represents the dependent form</comment>
<tRule>$aa=\uE03E;</tRule>
<tRule>$i=\uE03F;</tRule>
<tRule>$ii=\uE040;</tRule>
<tRule>$u=\uE041;</tRule>
<tRule>$uu=\uE042;</tRule>
<tRule>$rh=\uE043;</tRule>
<tRule>$rrh=\uE044;</tRule>
<tRule>$ce=\uE045; #VOWEL SIGN CANDRA E</tRule>
<tRule>$se=\uE046; #VOWEL SIGN SHORT E</tRule>
<tRule>$e=\uE047;</tRule>
<tRule>$ai=\uE048;</tRule>
<tRule>$co=\uE049; # VOWEL SIGN CANDRA O</tRule>
<tRule>$so=\uE04A; # VOWEL SIGN SHORT O</tRule>
<tRule>$o=\uE04B; # ो</tRule>
<tRule>$au=\uE04C;</tRule>
<tRule>$virama=\uE04D;</tRule>
<comment># \u094E Reserved</comment>
<comment># \u094F Reserved</comment>
<tRule>$om = \uE050; # OM</tRule>
<comment># ॑→; # UNMAPPED STRESS SIGN UDATTA</comment>
<comment># ॒→; # UNMAPPED STRESS SIGN ANUDATTA</comment>
<comment># ॓→; # UNMAPPED GRAVE ACCENT</comment>
<comment># ॔→; # UNMAPPED ACUTE ACCENT</comment>
<tRule>$lm = \uE055;# Telugu Length Mark</tRule>
<tRule>$ailm=\uE056;# AI Length Mark</tRule>
<tRule>$aulm=\uE057;# AU Length Mark</tRule>
<comment>#urdu compatibity forms</comment>
<tRule>$uka=\uE058;</tRule>
<tRule>$ukha=\uE059;</tRule>
<tRule>$ugha=\uE05A;</tRule>
<tRule>$ujha=\uE05B;</tRule>
<tRule>$uddha=\uE05C;</tRule>
<tRule>$udha=\uE05D;</tRule>
<tRule>$ufa=\uE05E;</tRule>
<tRule>$uya=\uE05F;</tRule>
<tRule>$wrr=\uE060;</tRule>
<tRule>$wll=\uE061;</tRule>
<tRule>$lh=\uE062;</tRule>
<tRule>$llh=\uE063;</tRule>
<tRule>$danda=\uE064;</tRule>
<tRule>$doubleDanda=\uE065;</tRule>
<tRule>$zero=\uE066; # DIGIT ZERO</tRule>
<tRule>$one=\uE067; # DIGIT ONE</tRule>
<tRule>$two=\uE068; # DIGIT TWO</tRule>
<tRule>$three=\uE069; # DIGIT THREE</tRule>
<tRule>$four=\uE06A; # DIGIT FOUR</tRule>
<tRule>$five=\uE06B; # DIGIT FIVE</tRule>
<tRule>$six=\uE06C; # DIGIT SIX</tRule>
<tRule>$seven=\uE06D; # DIGIT SEVEN</tRule>
<tRule>$eight=\uE06E; # DIGIT EIGHT</tRule>
<tRule>$nine=\uE06F; # DIGIT NINE</tRule>
<tRule>$dgs=\uE082;</tRule>
<comment># For all other scripts</comment>
<tRule>$ecp0=\uE070;</tRule>
<tRule>$ecp1=\uE071;</tRule>
<tRule>$ecp2=\uE072;</tRule>
<tRule>$ecp3=\uE073;</tRule>
<tRule>$ecp4=\uE074;</tRule>
<tRule>$ecp5=\uE075;</tRule>
<tRule>$ecp6=\uE076;</tRule>
<tRule>$ecp7=\uE077;</tRule>
<tRule>$ecp8=\uE078;</tRule>
<tRule>$ecp9=\uE079;</tRule>
<tRule>$ecpA=\uE07A;</tRule>
<tRule>$ecpB=\uE07B;</tRule>
<tRule>$ecpC=\uE07C;</tRule>
<tRule>$ecpD=\uE07D;</tRule>
<tRule>$ecpE=\uE07E;</tRule>
<tRule>$ecpF=\uE07F;</tRule>
<comment># Khanda-ta</comment>
<tRule>$kta=\uE083;</tRule>
<comment># ॰→; # nothing in Latin maps to InterIndic ABBREVIATION SIGN</comment>
<tRule>$depVowelAbove=[\uE03E-\uE040\uE045-\uE04C];</tRule>
<tRule>$depVowelBelow=[\uE041-\uE044];</tRule>
<tRule>$endThing=[$danda$doubleDanda];</tRule>
<comment># $x was originally called '§'; $z was '%'</comment>
<tRule>$x=[$virama$aa$ai$au$ii$i$uu$u$rrh$rh$lh$e$o$se$ce$so$co];</tRule>
<tRule>$z=[bcdfghjklmnpqrstvwxyz];</tRule>
<tRule>$consonants=[[$ka-$ha]$z[क-ह][ক-হ][ਕ-ਹ][ક-હ][କ-ହ][க-ஹ][క-హ][ಕ-ಹ][ക-ഹ]];</tRule>
<tRule>̕ → $avagraha;</tRule>
<tRule>̃→$chandrabindu$anusvara;</tRule>
<tRule>m̐→$chandrabindu;</tRule>
<tRule>ḥ→$visarga;</tRule>
<tRule>x→$ka$virama$sa;</tRule>
<comment># convert to independent forms at start of word or syllable:</comment>
<comment># dependent forms for roundtrip</comment>
<tRule>̔ā→$aa;</tRule>
<tRule>̔ai→$ai;</tRule>
<tRule>̔au→$au;</tRule>
<tRule>̔ii→$ii;</tRule>
<tRule>̔ī→$ii;</tRule>
<tRule>̔i→$i;</tRule>
<tRule>̔ū→$uu;</tRule>
<tRule>̔u→$u;</tRule>
<tRule>̔r̥̄→$rrh;</tRule>
<tRule>̔r̥→$rh;</tRule>
<tRule>̔l̥̄→$llh;</tRule>
<tRule>̔lh→$lh;</tRule>
<tRule>̔l̥→$lh;</tRule>
<tRule>̔ē→$e;</tRule>
<tRule>̔ō→$o;</tRule>
<tRule>̔a→;</tRule>
<tRule>̔ĕ→$ce;</tRule>
<tRule>̔ŏ→$co;</tRule>
<tRule>̔e→$se;</tRule>
<tRule>̔o→$so;</tRule>
<comment># preceeded by consonants</comment>
<tRule>$consonants{ ā→$aa;</tRule>
<tRule>$consonants{ ai→$ai;</tRule>
<tRule>$consonants{ au→$au;</tRule>
<tRule>$consonants{ ii→$ii;</tRule>
<tRule>$consonants{ ī→$ii;</tRule>
<tRule>$consonants{ i→$i;</tRule>
<tRule>$consonants{ ū→$uu;</tRule>
<tRule>$consonants{ u→$u;</tRule>
<tRule>$consonants{ r̥̄→$rrh;</tRule>
<tRule>$consonants{ r̥a→$rh;</tRule>
<tRule>$consonants{ r̥→$rh;</tRule>
<tRule>$consonants{ l̥̄→$llh;</tRule>
<tRule>$consonants{ lh→$lh;</tRule>
<tRule>$consonants{ l̥→$lh;</tRule>
<tRule>$consonants{ ē→$e;</tRule>
<tRule>$consonants{ ō→$o;</tRule>
<tRule>$consonants{ ĕ→$ce;</tRule>
<tRule>$consonants{ ŏ→$co;</tRule>
<tRule>$consonants{ e→$se;</tRule>
<tRule>$consonants{ o→$so;</tRule>
<comment># e.g. keai -→ {ka}{e}{wai}; k'ai -→ {ka}{wai}; (ai) -→ ({wai})</comment>
<tRule>ā→$waa;</tRule>
<tRule>ai→$wai;</tRule>
<tRule>au→$wau;</tRule>
<tRule>ī→$wii;</tRule>
<tRule>i→$wi;</tRule>
<tRule>ū→$wuu;</tRule>
<tRule>u→$wu;</tRule>
<tRule>r̥̄→$wrr;</tRule>
<tRule>r̥→$wr;</tRule>
<tRule>l̥̄→$wll;</tRule>
<tRule>lh→$wl;</tRule>
<tRule>l̥→$wl;</tRule>
<tRule>ē→$we;</tRule>
<tRule>ō→$wo;</tRule>
<tRule>a→$wa;</tRule>
<tRule>ĕ→$wce;</tRule>
<tRule>ŏ→$wco;</tRule>
<tRule>e→$wse;</tRule>
<tRule>''om→$om;</tRule>
<tRule>o→$wso;</tRule>
<comment># rules for anusvara</comment>
<tRule>n}r̥ → $na|$virama;</tRule>
<tRule>n}l̥ → $na|$virama;</tRule>
<tRule>n}na → $na|$virama;</tRule>
<tRule>ṅ}[kg] → $anusvara;</tRule>
<tRule>ṅ}ṅ → $anusvara;</tRule>
<tRule>n̄}[cj] → $anusvara;</tRule>
<tRule>n̄}ñ → $anusvara;</tRule>
<tRule>ṇ}[tdn]̣ → $anusvara;</tRule>
<tRule>n}[tdn] → $anusvara;</tRule>
<tRule>m}[pbm] → $anusvara;</tRule>
<tRule>n}[ylvshr] → $anusvara;</tRule>
<tRule>ṁ → $anusvara;</tRule>
<comment>#urdu compatibility</comment>
<tRule>q→$uka|$virama;</tRule>
<tRule>ḵẖ→$ukha |$virama;</tRule>
<tRule>ġ→ $ugha | $virama;</tRule>
<tRule>z → $ujha |$virama;</tRule>
<tRule>f → $ufa|$virama;</tRule>
<tRule>ṯ→$kta;</tRule>
<comment># dev</comment>
<tRule>ẏ→$uya|$virama;</tRule>
<tRule>ḻ→$ela|$virama;</tRule>
<tRule>ṉ→$ena|$virama;</tRule>
<tRule>ṅ→$nga|$virama;</tRule>
<tRule>ñ→$nya|$virama;</tRule>
<tRule>ṇ→$nna|$virama;</tRule>
<tRule>ṭh→$ttha|$virama;</tRule>
<tRule>ṭ→$tta|$virama;</tRule>
<tRule>ṛh→$udha|$virama;</tRule>
<tRule>ṛ→$uddha|$virama;</tRule>
<tRule>ḍh→$ddha|$virama;</tRule>
<tRule>ḍ→$dda|$virama;</tRule>
<tRule>kh→$kha|$virama;</tRule>
<tRule>k→$ka|$virama;</tRule>
<tRule>gh→$gha|$virama;</tRule>
<tRule>g→$ga|$virama;</tRule>
<tRule>ch→$cha|$virama;</tRule>
<tRule>c→$ca|$virama;</tRule>
<tRule>jh→$jha|$virama;</tRule>
<tRule>j→$ja|$virama;</tRule>
<tRule>ny→$nya|$virama;</tRule>
<tRule>tth→$ttha|$virama;</tRule>
<tRule>ddh→$ddha|$virama;</tRule>
<tRule>th→$tha|$virama;</tRule>
<tRule>t→$ta|$virama;</tRule>
<tRule>dh→$dha|$virama;</tRule>
<tRule>d→$da|$virama;</tRule>
<tRule>n→$na|$virama;</tRule>
<tRule>ph→$pha|$virama;</tRule>
<tRule>p→$pa|$virama;</tRule>
<tRule>bh→$bha|$virama;</tRule>
<tRule>b→$ba|$virama;</tRule>
<tRule>m→$ma|$virama;</tRule>
<tRule>y→$ya|$virama;</tRule>
<tRule>ṟ→$rra|$virama;</tRule>
<tRule>r→$ra|$virama;</tRule>
<tRule>ḷ→$lla|$virama;</tRule>
<tRule>l→$la|$virama;</tRule>
<tRule>v→$va|$virama;</tRule>
<tRule>ẇ→$vva|$virama;</tRule>
<tRule>w→$va|$virama;</tRule>
<tRule>sh→$sha|$virama;</tRule>
<tRule>ss→$ssa|$virama;</tRule>
<tRule>ṣ→$ssa|$virama;</tRule>
<tRule>ś→$sha|$virama;</tRule>
<tRule>s→$sa|$virama;</tRule>
<tRule>h→$ha|$virama;</tRule>
<tRule>'.'→$danda;</tRule>
<tRule>$danda'.'→$doubleDanda;</tRule>
<tRule>$depVowelAbove{'~'→$anusvara;</tRule>
<tRule>$depVowelBelow{'~'→$chandrabindu;</tRule>
<comment># convert to dependent forms after consonant with no vowel:</comment>
<comment># e.g. kai -→ {ka}{virama}ai -→ {ka}{ai}</comment>
<comment>#$virama aa→$aa;</comment>
<tRule>$virama ā→$aa;</tRule>
<tRule>$virama ai→$ai;</tRule>
<tRule>$virama au→$au;</tRule>
<tRule>$virama ii→$ii;</tRule>
<tRule>$virama ī→$ii;</tRule>
<tRule>$virama i→$i;</tRule>
<comment>#$virama uu→$uu;</comment>
<tRule>$virama ū→$uu;</tRule>
<tRule>$virama u→$u;</tRule>
<comment>#$virama rrh→$rrh;</comment>
<tRule>$virama r̥̄→$rrh;</tRule>
<comment>#$virama rh→$rh;</comment>
<tRule>$virama r̥a→$rh;</tRule>
<tRule>$virama r̥→$rh;</tRule>
<tRule>$virama l̥̄→$llh;</tRule>
<tRule>$virama lh→$lh;</tRule>
<tRule>$virama l̥→$lh;</tRule>
<tRule>$virama ē→$e;</tRule>
<tRule>$virama ō→$o;</tRule>
<tRule>$virama a→;</tRule>
<tRule>$virama ĕ→$ce;</tRule>
<tRule>$virama ŏ→$co;</tRule>
<tRule>$virama e→$se;</tRule>
<tRule>$virama o→$so;</tRule>
<comment># otherwise convert independent forms when separated by ': k'ai -→ {ka}{virama}{wai}</comment>
<comment>#$virama''aa→$waa;</comment>
<tRule>$virama''ā→$waa;</tRule>
<tRule>$virama''ai→$wai;</tRule>
<tRule>$virama''au→$wau;</tRule>
<comment>#$virama''ii→$wii;</comment>
<tRule>$virama''ī→$wii;</tRule>
<tRule>$virama''i→$wi;</tRule>
<comment>#$virama''uu→$wuu;</comment>
<tRule>$virama''ū→$wuu;</tRule>
<tRule>$virama''u→$wu;</tRule>
<comment>#$virama''rrh→$wrr;</comment>
<tRule>$virama''r̥̄→$wrr;</tRule>
<comment>#$virama''rh→$wr;</comment>
<tRule>$virama''r̥→$wr;</tRule>
<tRule>$virama''l̥̄→$wll;</tRule>
<comment>#$virama''lh→$wl;</comment>
<tRule>$virama''l̥→$wl;</tRule>
<tRule>$virama''ē→$we;</tRule>
<tRule>$virama''ō→$wo;</tRule>
<tRule>$virama''a→$wa;</tRule>
<tRule>$virama''ĕ→$wce;</tRule>
<tRule>$virama''ŏ→$wco;</tRule>
<tRule>$virama''e→$wse;</tRule>
<tRule>$virama''o→$wso;</tRule>
<comment># no virama</comment>
<tRule>''ā→$waa;</tRule>
<tRule>''ai→$wai;</tRule>
<tRule>''au→$wau;</tRule>
<tRule>''ī→$wii;</tRule>
<tRule>''i→$wi;</tRule>
<tRule>''ū→$wuu;</tRule>
<tRule>''u→$wu;</tRule>
<tRule>''r̥̄→$wrr;</tRule>
<tRule>''r̥→$wr;</tRule>
<tRule>''l̥̄→$wll;</tRule>
<tRule>''l̥→$wl;</tRule>
<tRule>''ē→$we;</tRule>
<tRule>''ō→$wo;</tRule>
<tRule>''a→$wa;</tRule>
<tRule>''ĕ→$wce;</tRule>
<tRule>''ŏ→$wco;</tRule>
<tRule>''e→$wse;</tRule>
<tRule>''o→$wso;</tRule>
<tRule>$virama } [$z] → $virama;</tRule>
<tRule>$virama } ' ' → $virama ;</tRule>
<tRule>$virama}$endThing→;</tRule>
<tRule>ʔ→$dgs; # Glottal Stop</tRule>
<tRule>0→$zero;</tRule>
<tRule>1→$one;</tRule>
<tRule>2→$two;</tRule>
<tRule>3→$three;</tRule>
<tRule>4→$four;</tRule>
<tRule>5→$five;</tRule>
<tRule>6→$six;</tRule>
<tRule>7→$seven;</tRule>
<tRule>8→$eight;</tRule>
<tRule>9→$nine;</tRule>
<tRule>''→;</tRule>
<comment>#:: NFC (NFD) ;</comment>
</transform>
</transforms>
</supplementalData>