git.saurik.com Git - apple/icu.git/blame_incremental - icuSources/data/translit/zh_Latn_PINYIN

... / ...

Commit	Line	Data
	1	# ***************************************************************************
	2	# *
	3	# * Copyright (C) 2004-2016, International Business Machines
	4	# * Corporation; Unicode, Inc.; and others. All Rights Reserved.
	5	# *
	6	# ***************************************************************************
	7	# File: zh_Latn_PINYIN_ru.txt
	8	# Generated from CLDR
	9	#
	10
	11	# Cyrillization of Mandarin Chinese from Pinyin into Russian (Palladius system).
	12	#
	13	# References:
	14	# http://ru.wikipedia.org/wiki/Транскрипционная_система_Палладия
	15	# http://www.omniglot.com/writing/mandarin_pts.htm
	16	# http://www.pinyin.info/romanization/russian/index.html
	17	# These differ in the treatment of some syllables (e.g. <jue>) from the rules
	18	# below.
	19	#
	20	# Further commentary:
	21	# http://languagelog.ldc.upenn.edu/nll/?p=604
	22	#
	23	# Remove tone marks.
	24	:: NFD (NFC);
	25	[\u0304\u0301\u030C\u0300\u0306] → ;
	26	:: NFC (NFD);
	27	#
	28	#
	29	# Syllabify. Add apostrophes to disambiguate whether <n> and <g> belong to the
	30	# coda or onset of a syllable.
	31	:: Null ();
	32	ng } [aeou] → n\'g;
	33	ng → ng\';
	34	n } [aeiouü] → \'n;
	35	n → n\';
	36	#
	37	#
	38	# Main pass.
	39	:: Null ();
	40	#
	41	#
	42	Ai → Ай;
	43	A → А;
	44	B → Б;
	45	Ch → Ч;
	46	Ci → Цы;
	47	C → Ц;
	48	D → Д;
	49	E → Э;
	50	F → Ф;
	51	G → Г;
	52	H → Х;
	53	Ju → Цз \| ü;
	54	J → Цз;
	55	K → К;
	56	L → Л;
	57	M → М;
	58	N → Н;
	59	O → О;
	60	P → П;
	61	Qu → Ц \| ü;
	62	Q → Ц;
	63	R → Ж;
	64	Sh → Ш;
	65	Si → Сы;
	66	S → С;
	67	T → Т;
	68	Wu → У;
	69	W → В;
	70	Xu → С \| ü;
	71	X → С;
	72	Yai → Яй;
	73	Ya → Я;
	74	Ye → Е;
	75	Yi → И;
	76	You → Ю;
	77	Yo } ng → Ю;
	78	Yo → Ё;
	79	Yu → Ю \| v;
	80	Zh → Чж;
	81	Zi → Цзы;
	82	Z → Цз;
	83	#
	84	#
	85	[$] { n → н;
	86	\'n → н;
	87	\' → ;
	88	#
	89	#
	90	ai → ай;
	91	a → а;
	92	b → б;
	93	ch → ч;
	94	ci → цы;
	95	c → ц;
	96	d → д;
	97	ei → эй;
	98	e → э;
	99	f → ф;
	100	g → г;
	101	h → х;
	102	ia → я;
	103	ie → е;
	104	io } ng → ю;
	105	iu → ю;
	106	i → и;
	107	ju → цз \| ü;
	108	j → цз;
	109	k → к;
	110	l → л;
	111	m → м;
	112	ng → н;
	113	n → нь;
	114	o } ng → у;
	115	o → о;
	116	p → п;
	117	qu → ц \| ü;
	118	q → ц;
	119	r } [^aeiou] → р;
	120	r → ж;
	121	sh → ш;
	122	si → сы;
	123	s → с;
	124	t → т;
	125	[Хх] { ui → уэй; # special exception
	126	ui → уй;
	127	uo → о;
	128	u → у;
	129	ü → ю \| v;
	130	#
	131	# We use the dummy symbol <v> to signal that an <ü> or equivalent character has
	132	# just been processed.
	133	ve → э;
	134	#
	135	#alternative: ve → е;
	136	v } [an] → ;
	137	v → й;
	138	wu → у;
	139	w → в;
	140	xu → с \| ü;
	141	x → с;
	142	yai → яй;
	143	ya → я;
	144	ye → е;
	145	yi → и;
	146	you → ю;
	147	yo } ng → ю;
	148	yo → ё;
	149	yu → ю \| v;
	150	zh → чж;
	151	zi → цзы;
	152	z → цз;
	153	#
	154	#
	155	# Final pass: Make sure that the output consists entirely of Cyrillic letters.
	156	:: NFC ();
	157