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

Commit	Line	Data
f3c0d7a5 A	1	# © 2016 and later: Unicode, Inc. and others.
	2	# License & terms of use: http://www.unicode.org/copyright.html#License
	3	#
729e4ab9	4	# File: zh_Latn_PINYIN_ru.txt
f3c0d7a5	5	# Generated from CLDR
729e4ab9	6	#
2ca993e8 A	7
	8	# Cyrillization of Mandarin Chinese from Pinyin into Russian (Palladius system).
	9	#
	10	# References:
	11	# http://ru.wikipedia.org/wiki/Транскрипционная_система_Палладия
	12	# http://www.omniglot.com/writing/mandarin_pts.htm
	13	# http://www.pinyin.info/romanization/russian/index.html
	14	# These differ in the treatment of some syllables (e.g. <jue>) from the rules
	15	# below.
	16	#
	17	# Further commentary:
	18	# http://languagelog.ldc.upenn.edu/nll/?p=604
	19	#
	20	# Remove tone marks.
729e4ab9 A	21	:: NFD (NFC);
	22	[\u0304\u0301\u030C\u0300\u0306] → ;
	23	:: NFC (NFD);
2ca993e8 A	24	#
	25	#
	26	# Syllabify. Add apostrophes to disambiguate whether <n> and <g> belong to the
	27	# coda or onset of a syllable.
729e4ab9 A	28	:: Null ();
	29	ng } [aeou] → n\'g;
	30	ng → ng\';
	31	n } [aeiouü] → \'n;
	32	n → n\';
2ca993e8 A	33	#
	34	#
	35	# Main pass.
729e4ab9	36	:: Null ();
2ca993e8 A	37	#
2ca993e8 A	38	#
729e4ab9	39	Ai → Ай;
51004dcb A	40	A → А;
51004dcb A	41	B → Б;
729e4ab9 A	42	Ch → Ч;
729e4ab9 A	43	Ci → Цы;
51004dcb A	44	C → Ц;
	45	D → Д;
	46	E → Э;
	47	F → Ф;
	48	G → Г;
	49	H → Х;
729e4ab9	50	Ju → Цз \| ü;
51004dcb A	51	J → Цз;
	52	K → К;
	53	L → Л;
	54	M → М;
	55	N → Н;
	56	O → О;
	57	P → П;
729e4ab9	58	Qu → Ц \| ü;
51004dcb A	59	Q → Ц;
51004dcb A	60	R → Ж;
729e4ab9 A	61	Sh → Ш;
729e4ab9 A	62	Si → Сы;
51004dcb A	63	S → С;
51004dcb A	64	T → Т;
729e4ab9	65	Wu → У;
51004dcb	66	W → В;
729e4ab9	67	Xu → С \| ü;
51004dcb	68	X → С;
729e4ab9 A	69	Yai → Яй;
	70	Ya → Я;
	71	Ye → Е;
	72	Yi → И;
	73	You → Ю;
	74	Yo } ng → Ю;
	75	Yo → Ё;
	76	Yu → Ю \| v;
	77	Zh → Чж;
	78	Zi → Цзы;
51004dcb	79	Z → Цз;
2ca993e8 A	80	#
2ca993e8 A	81	#
729e4ab9 A	82	[$] { n → н;
729e4ab9 A	83	\'n → н;
51004dcb	84	\' → ;
2ca993e8 A	85	#
2ca993e8 A	86	#
729e4ab9	87	ai → ай;
51004dcb A	88	a → а;
51004dcb A	89	b → б;
729e4ab9 A	90	ch → ч;
729e4ab9 A	91	ci → цы;
51004dcb A	92	c → ц;
51004dcb A	93	d → д;
729e4ab9	94	ei → эй;
51004dcb A	95	e → э;
	96	f → ф;
	97	g → г;
	98	h → х;
729e4ab9 A	99	ia → я;
	100	ie → е;
	101	io } ng → ю;
	102	iu → ю;
51004dcb	103	i → и;
729e4ab9	104	ju → цз \| ü;
51004dcb A	105	j → цз;
	106	k → к;
	107	l → л;
	108	m → м;
729e4ab9	109	ng → н;
51004dcb	110	n → нь;
729e4ab9	111	o } ng → у;
51004dcb A	112	o → о;
51004dcb A	113	p → п;
729e4ab9	114	qu → ц \| ü;
51004dcb	115	q → ц;
729e4ab9	116	r } [^aeiou] → р;
51004dcb	117	r → ж;
729e4ab9 A	118	sh → ш;
729e4ab9 A	119	si → сы;
51004dcb A	120	s → с;
	121	t → т;
	122	[Хх] { ui → уэй; # special exception
729e4ab9 A	123	ui → уй;
729e4ab9 A	124	uo → о;
51004dcb A	125	u → у;
51004dcb A	126	ü → ю \| v;
2ca993e8 A	127	#
	128	# We use the dummy symbol <v> to signal that an <ü> or equivalent character has
	129	# just been processed.
729e4ab9	130	ve → э;
2ca993e8 A	131	#
2ca993e8 A	132	#alternative: ve → е;
729e4ab9	133	v } [an] → ;
51004dcb	134	v → й;
729e4ab9	135	wu → у;
51004dcb	136	w → в;
729e4ab9	137	xu → с \| ü;
51004dcb	138	x → с;
729e4ab9 A	139	yai → яй;
	140	ya → я;
	141	ye → е;
	142	yi → и;
	143	you → ю;
	144	yo } ng → ю;
	145	yo → ё;
	146	yu → ю \| v;
	147	zh → чж;
	148	zi → цзы;
51004dcb	149	z → цз;
2ca993e8 A	150	#
	151	#
	152	# Final pass: Make sure that the output consists entirely of Cyrillic letters.
729e4ab9	153	:: NFC ();
2ca993e8	154