larvitgeodata
Version:
Geo data, primarily ISO territories, languages etc. Data fetched mostly from CLDR.
174 lines (173 loc) • 5.29 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="zh_Latn_PINYIN" target="ru" direction="forward">
<comment>
# Copyright (C) 2009 Google, Inc. All Rights Reserved.
# Author: mjansche@google.com (Martin Jansche)
#
# Cyrillization of Mandarin Chinese from Pinyin into Russian (Palladius system).
#
# References:
# http://ru.wikipedia.org/wiki/Транскрипционная_система_Палладия
# http://www.omniglot.com/writing/mandarin_pts.htm
# http://www.pinyin.info/romanization/russian/index.html
# These differ in the treatment of some syllables (e.g. <jue>) from the rules
# below.
#
# Further commentary:
# http://languagelog.ldc.upenn.edu/nll/?p=604
#
# Remove tone marks.
</comment>
<tRule>:: NFD (NFC);</tRule>
<tRule>[\u0304\u0301\u030C\u0300\u0306] → ;</tRule>
<tRule>:: NFC (NFD);</tRule>
<comment>
#
# Syllabify. Add apostrophes to disambiguate whether <n> and <g> belong to the
# coda or onset of a syllable.
</comment>
<tRule>:: Null ();</tRule>
<tRule>ng } [aeou] → n\'g;</tRule>
<tRule>ng → ng\';</tRule>
<tRule>n } [aeiouü] → \'n;</tRule>
<tRule>n → n\';</tRule>
<comment>
#
# Main pass.
</comment>
<tRule>:: Null ();</tRule>
<comment>
#
</comment>
<tRule>Ai → Ай;</tRule>
<tRule>A → А;</tRule>
<tRule>B → Б;</tRule>
<tRule>Ch → Ч;</tRule>
<tRule>Ci → Цы;</tRule>
<tRule>C → Ц;</tRule>
<tRule>D → Д;</tRule>
<tRule>E → Э;</tRule>
<tRule>F → Ф;</tRule>
<tRule>G → Г;</tRule>
<tRule>H → Х;</tRule>
<tRule>Ju → Цз | ü;</tRule>
<tRule>J → Цз;</tRule>
<tRule>K → К;</tRule>
<tRule>L → Л;</tRule>
<tRule>M → М;</tRule>
<tRule>N → Н;</tRule>
<tRule>O → О;</tRule>
<tRule>P → П;</tRule>
<tRule>Qu → Ц | ü;</tRule>
<tRule>Q → Ц;</tRule>
<tRule>R → Ж;</tRule>
<tRule>Sh → Ш;</tRule>
<tRule>Si → Сы;</tRule>
<tRule>S → С;</tRule>
<tRule>T → Т;</tRule>
<tRule>Wu → У;</tRule>
<tRule>W → В;</tRule>
<tRule>Xu → С | ü;</tRule>
<tRule>X → С;</tRule>
<tRule>Yai → Яй;</tRule>
<tRule>Ya → Я;</tRule>
<tRule>Ye → Е;</tRule>
<tRule>Yi → И;</tRule>
<tRule>You → Ю;</tRule>
<tRule>Yo } ng → Ю;</tRule>
<tRule>Yo → Ё;</tRule>
<tRule>Yu → Ю | v;</tRule>
<tRule>Zh → Чж;</tRule>
<tRule>Zi → Цзы;</tRule>
<tRule>Z → Цз;</tRule>
<comment>
#
</comment>
<tRule>[$] { n → н;</tRule>
<tRule>\'n → н;</tRule>
<tRule>\' → ;</tRule>
<comment>
#
</comment>
<tRule>ai → ай;</tRule>
<tRule>a → а;</tRule>
<tRule>b → б;</tRule>
<tRule>ch → ч;</tRule>
<tRule>ci → цы;</tRule>
<tRule>c → ц;</tRule>
<tRule>d → д;</tRule>
<tRule>ei → эй;</tRule>
<tRule>e → э;</tRule>
<tRule>f → ф;</tRule>
<tRule>g → г;</tRule>
<tRule>h → х;</tRule>
<tRule>ia → я;</tRule>
<tRule>ie → е;</tRule>
<tRule>io } ng → ю;</tRule>
<tRule>iu → ю;</tRule>
<tRule>i → и;</tRule>
<tRule>ju → цз | ü;</tRule>
<tRule>j → цз;</tRule>
<tRule>k → к;</tRule>
<tRule>l → л;</tRule>
<tRule>m → м;</tRule>
<tRule>ng → н;</tRule>
<tRule>n → нь;</tRule>
<tRule>o } ng → у;</tRule>
<tRule>o → о;</tRule>
<tRule>p → п;</tRule>
<tRule>qu → ц | ü;</tRule>
<tRule>q → ц;</tRule>
<tRule>r } [^aeiou] → р;</tRule>
<tRule>r → ж;</tRule>
<tRule>sh → ш;</tRule>
<tRule>si → сы;</tRule>
<tRule>s → с;</tRule>
<tRule>t → т;</tRule>
<tRule>[Хх] { ui → уэй; # special exception</tRule>
<tRule>ui → уй;</tRule>
<tRule>uo → о;</tRule>
<tRule>u → у;</tRule>
<tRule>ü → ю | v;</tRule>
<comment>
# We use the dummy symbol <v> to signal that an <ü> or equivalent character has
# just been processed.
</comment>
<tRule>ve → э;</tRule>
<comment>
#alternative: ve → е;
</comment>
<tRule>v } [an] → ;</tRule>
<tRule>v → й;</tRule>
<tRule>wu → у;</tRule>
<tRule>w → в;</tRule>
<tRule>xu → с | ü;</tRule>
<tRule>x → с;</tRule>
<tRule>yai → яй;</tRule>
<tRule>ya → я;</tRule>
<tRule>ye → е;</tRule>
<tRule>yi → и;</tRule>
<tRule>you → ю;</tRule>
<tRule>yo } ng → ю;</tRule>
<tRule>yo → ё;</tRule>
<tRule>yu → ю | v;</tRule>
<tRule>zh → чж;</tRule>
<tRule>zi → цзы;</tRule>
<tRule>z → цз;</tRule>
<comment>
#
# Final pass: Make sure that the output consists entirely of Cyrillic letters.
</comment>
<tRule>:: NFC ();</tRule>
</transform>
</transforms>
</supplementalData>