cldr-transforms 46.0.0-BETA2

package/transforms/pl-pl_FONIPA.txt

@@ -0,0 +1,113 @@
+
+# Polish orthography to phonemic transcription.
+# http://en.wikipedia.org/wiki/Polish_phonology
+#
+# Transform input to normalized form NFC, and to lowercase.
+:: NFC () ;
+:: Lower () ;
+#
+#
+# Definitions.
+$voiceless = [cfhkpst];
+$vowel = [ aąeęioóuy ];
+#
+#
+# Digraphs and Trigraphs.
+#
+ch } i → ç ;
+ch → x ;
+ci } $vowel → t \u0361 ɕ ;
+ci → t \u0361 ɕ i ;
+cz → t \u0361 ʂ ;
+dzi } $vowel → d \u0361 ʑ ;
+dzi → d \u0361 ʑ i ;
+dz } $voiceless → t \u0361 s ;
+dz → d \u0361 z ;
+dź } $voiceless → t \u0361 ɕ ;
+dź → d \u0361 ʑ ;
+dż } $voiceless → t \u0361 ʂ ;
+dż → d \u0361 ʐ ;
+kw → kf ;
+krw → krf ;
+ni } $vowel → ɲ ;
+ni → ɲ i ;
+$voiceless { rz → ʂ ;
+rz } $voiceless → ʂ ;
+rz → ʐ ;
+sz → ʂ ;
+trw → trf ;
+tw → tf ;
+zi } $vowel → ʑ ;
+zi → ʑ i ;
+#
+#
+a → a ;
+ą } [bp] → ɔm ;
+ą } [kg] → ɔŋ ;
+ą } [cdt] → ɔn ;
+ą → ɔ̃ ;
+b } $voiceless → p ;
+b → b ;
+c → t \u0361 s ;
+ć → t \u0361 ɕ ;
+d } $voiceless → t ;
+d → d ;
+e → ɛ ;
+ę } [bp] → ɛm ;
+ę } [dt] → ɛn ;
+ę } [gk] → ɛŋ ;
+ę → ɛ̃;
+f → f ;
+g } $voiceless → k ;
+g → ɡ ;
+h } i → ç ;
+h → x ;
+i } $vowel → ʲ ;
+i → i ;
+j → j ;
+k } [bdzż] → ɡ ;
+k → k ;
+l } i → ʎ ;
+l → l ;
+ł → w ;
+m → m ;
+n → n ;
+ń → ɲ ;
+o → ɔ ;
+ó → u ;
+r → r ;
+si } $vowel → ɕ ;
+si → ɕi ;
+s → s ;
+ś → ɕ ;
+t → t ;
+u → u ;
+w } $voiceless → f ;
+w → v ;
+y → ɨ ;
+ź } $voiceless → ɕ ;
+ź → ʑ ;
+ż } $voiceless → ʂ ;
+ż → ʐ ;
+#
+#
+# Second pass: Phoneme-to-phone rules.
+:: Null ;
+#
+#
+tʐ → tʂ ;
+pʐ → pʂ ;
+xʐ → xʂ ;
+ʐt͡ʂ → ʂt͡ʂ ;
+#
+#
+# Final de-voicing.
+b } [$] → p ;
+d \u0361 z } [$] → t \u0361 s ;  # rydz → rɨt͡s
+d } [$] → t ;
+ɡ } [$] → k ;
+v } [$] → f ;
+ʐ } [$] → ʂ ;
+ʑ } [$] → ɕ ;
+z } [$] → s ;
+

package/transforms/pl_FONIPA-ja.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "ja-t-pl-fonipa",
+  "_source": "pl_FONIPA",
+  "_direction": "forward",
+  "_target": "ja",
+  "_rulesFile": "pl_FONIPA-ja.txt"
+}

package/transforms/pl_FONIPA-ja.txt

@@ -0,0 +1,301 @@
+
+# Transforms a Phonemic IPA transcription of Polish (pl_FONIPA) to Katakana.
+#
+$word_boundary = [-\ $] ;
+$vowel = [aeiouw] ;       # Vowels and glides
+$not_vowel = [^$vowel] ;
+#
+#
+# First pass: Collapse phonetic distinctions not preserved in Katakana.
+ç → | h;
+ɡ → | g;
+ʎ → | l;
+ŋ → | n;
+d \u0361 ʑ → | ʑ;
+d \u0361 ʐ → | ʐ;
+d \u0361 z → | z;
+#
+#
+ɛ̃ → | en;
+ɛ → | e;
+[ɨʲ] → | i;
+ɔ̃ → | on;
+ɔ → | o;
+#
+#
+:: Null ();
+#
+#
+# Main pass: Phoneme to Katakana conversion.
+'.' → ;
+a → ア;
+#
+#
+ba → バ;
+bb → ッ | b;
+be → ベ;
+b[ij]a → ビャ;
+b[ij]o → ビョ;
+b[ij] → ビ;
+bo → ボ;
+bu → ブ;
+b } $word_boundary → プ;
+b → ブ;
+#
+#
+ca → チャ ;
+ce → チェ ;
+ci → チ ;
+cu → チュ ;
+co → チョ ;
+c → チ ;
+#
+#
+^ d \u0361 ɕ → dɕ;
+d \u0361 ɕ → ッ | dɕ;
+#
+#
+da → ダ;
+dd → ッ | d;
+de → デ;
+di → ディ;
+do → ド;
+du → ドゥ;
+dɕ → チ;
+d } $word_boundary → ト;
+d → ド;
+#
+#
+e → エ;
+#
+#
+fa → ファ;
+fe → フェ;
+ff → ッ | f;
+fi → フィ;
+fo → フォ;
+fu → フ;
+f → フ;
+#
+#
+ha → ハ;
+hi → ヒ;
+hu → フ;
+he → ヘ;
+ho → ホ;
+h } $word_boundary → ;
+h → フ;
+#
+#
+ga → ガ;
+ge → グエ;
+gi → ギ;
+gg → ッ | g;
+go → ゴ;
+gu → グ;
+g } $word_boundary → ク;
+g → グ;
+#
+#
+i → イ ;
+#
+#
+ja → ヤ;
+ji → イ;
+jo → ヨ;
+je → イェ;
+ju → ユ;
+j → イ;
+#
+#
+ka → カ;
+ke → ケ;
+ki → キ;
+kk → ッ | k;
+ko → コ;
+ku → ク;
+k → ク;
+#
+#
+la → ラ ;
+le → レ ;
+li → リ ;
+lho → ロ ;
+lo → ロ ;
+lu → ル ;
+l → ル ;
+#
+#
+ma → マ ;
+me → メ ;
+mi → ミ ;
+mo → モ ;
+mu → ム ;
+m } [bp] → ン ;
+m → ム ;
+#
+#
+na → ナ ;
+ne → ネ ;
+ni → ニ ;
+no → ノ ;
+nu → ヌ ;
+n → ン ;
+#
+#
+ɲa → ニャ ;
+ɲe → ニエ ;
+ɲi → ニ ;
+ɲo → ニョ ;
+ɲu → ニュ ;
+ɲ → ン ;
+#
+#
+o → オ ;
+#
+#
+pa → パ ;
+pe → ペ ;
+pio → ピョ ;
+pi → ピ ;
+po → ポ ;
+pp → ッ | p;
+pu → プ ;
+p → プ ;
+#
+#
+ra → ラ ;
+re → レ ;
+ri → リ ;
+ro → ロ ;
+ru → ル ;
+r → ル;
+#
+#
+sa → サ ;
+se → セ ;
+si → シ ;
+so → ソ ;
+su → ス ;
+s → ス ;
+#
+#
+ɕa → シャ; # not backed by data
+ɕe → シェ;
+ɕu → シュ; # not backed by data
+ɕo → ショ; # not backed by data
+ɕvi → シフィ;
+ɕi → シ;
+ɕ → シ;
+#
+#
+ʂa → シャ;
+ʂe → シェ;
+ʂu → シュ;
+ʂo → ショ; # not backed by data
+ʂi → シ;
+ʂ → シュ;
+#
+#
+#tʂa → ツァ;
+#tʂi → トシ;
+#tʂu → チュ;
+#tʂe → トシェ;
+#tʂ } $word_boundary → チ;
+#tʂ → チュ;
+#
+tɕa → チャ;
+tɕe → チェ;
+tɕi → チ;
+tɕu → チュ;
+tɕo → チョ;
+tɕ → チ;
+#
+#
+ta → タ;
+te → テ ;
+ti → ティ ;
+to → ト ;
+tu → トゥ ;
+#
+#
+tsa → ツァ ;
+tse → ツェ ;
+ts[ij] → ツィ ;
+tso → ツォ ;
+tsu → ツ ;
+ts → ツ ;
+#
+#
+^tt → ト | t;
+tt → ッ | t;
+#
+#
+t \u0361 ʂa → ツァ;
+t \u0361 ʂi → チ;
+t \u0361 ʂu → チュ;
+t \u0361 ʂe → チェ;
+t \u0361 ʂo → チョ;
+t \u0361 ʂ } $word_boundary → チ;
+t \u0361 ʂ → チュ;
+t \u0361 → | t;
+t → ト ;
+#
+#
+u → ウ ;
+#
+#
+va → バ;
+ve → ベ;
+vi → ビ;
+vo → ボ;
+vu → ブ;
+v } $word_boundary → フ;
+v → ブ;
+#
+#
+wa → ワ;
+wu → ウ;
+wi → ウィ;
+we → ウェ;
+wo → ウォ;
+w → ウ;
+#
+#
+xa → ハ ;
+xe → ヘ ;
+xi → ヒ ;
+xo → ホ ;
+xu → フ ;
+x → フ ;
+#
+#
+ʐa → ジャ;
+ʐe → ジェ;
+ʐi → ジ;
+ʐo → ジョ;
+ʐu → ジュ;
+ʐ } $word_boundary → ジュ;
+ʐ → ジ;
+#
+#
+ʑa → ジャ;
+ʑi → ジ;
+ʑo → ジオ;
+ʑe → ジェ;
+ʑu → ジュ;
+ʑ → ジ;
+#
+#
+za → ザ;
+ze → ジェ;
+zi → ジ;
+zo → ゾ;
+zu → ズ;
+z } $word_boundary → ス;
+z → ズ;
+#
+#
+#' ' → ・;
+#
+

package/transforms/rm_SURSILV-am.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "am-t-rm-sursilv",
+  "_source": "rm_SURSILV",
+  "_direction": "forward",
+  "_target": "am",
+  "_rulesFile": "rm_SURSILV-am.txt"
+}

package/transforms/rm_SURSILV-am.txt

@@ -0,0 +1,4 @@
+
+::rm_SURSILV-rm_FONIPA_SURSILV;
+::am_FONIPA-am;
+

package/transforms/rm_SURSILV-ar.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "ar-t-rm-sursilv",
+  "_source": "rm_SURSILV",
+  "_direction": "forward",
+  "_target": "ar",
+  "_rulesFile": "rm_SURSILV-ar.txt"
+}

package/transforms/rm_SURSILV-ar.txt

@@ -0,0 +1,4 @@
+
+::rm_SURSILV-rm_FONIPA_SURSILV;
+::und_FONIPA-ar;
+

package/transforms/rm_SURSILV-chr.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "chr-t-rm-sursilv",
+  "_source": "rm_SURSILV",
+  "_direction": "forward",
+  "_target": "chr",
+  "_rulesFile": "rm_SURSILV-chr.txt"
+}

package/transforms/rm_SURSILV-chr.txt

@@ -0,0 +1,4 @@
+
+::rm_SURSILV-rm_FONIPA_SURSILV;
+::und_FONIPA-chr;
+

package/transforms/rm_SURSILV-fa.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "fa-t-rm-sursilv",
+  "_source": "rm_SURSILV",
+  "_direction": "forward",
+  "_target": "fa",
+  "_rulesFile": "rm_SURSILV-fa.txt"
+}

package/transforms/rm_SURSILV-fa.txt

@@ -0,0 +1,4 @@
+
+::rm_SURSILV-rm_FONIPA_SURSILV;
+::und_FONIPA-fa;
+

package/transforms/rm_SURSILV-rm_FONIPA_SURSILV.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "rm-fonipa-sursilv-t-rm-sursilv",
+  "_source": "rm_SURSILV",
+  "_direction": "forward",
+  "_target": "rm_FONIPA_SURSILV",
+  "_rulesFile": "rm_SURSILV-rm_FONIPA_SURSILV.txt"
+}

package/transforms/rm_SURSILV-rm_FONIPA_SURSILV.txt

@@ -0,0 +1,96 @@
+
+# Transforms Rumantsch Sursilvan (rm-sursilv) to IPA (rm-fonipa-sursilv).
+#
+# References
+# ----------
+# [1] https://als.wikipedia.org/wiki/Surselvisch#Orthographi
+#     with links to sound recordings on Wikimedia Commons
+#
+# Output
+# ------
+# m n ɲ
+# p b t d c ɟ k ɡ
+# f v s z ʃ ʒ h
+# t͡ʃ t͡s
+# r l j ʎ
+# w
+# i u e ʊ ɛ ɔ a
+# ɪa̯ ɪa̯ʊ̯ ɪʊ̯ ɪɛ̯ u ʊa̯ ʊa̯ʊ̯ ʊɛ̯ ʊɛ̯ɪ̯ ʊɔ̯
+# ɛɪ̯ ɛʊ̯ aɪ̯ aʊ̯
+::Lower;
+ai → aɪ̯ ;
+au → aʊ̯ ;
+a → a ;
+b → b ;
+{c} [ei] → t͡s ;
+c → k ;
+d → d ;
+ei → ɛɪ̯ ;  # can also be /aɪ̯/ or /ɔɪ̯/ in some regions
+eu → ɛʊ̯ ;
+e → e ;  # can also be /ɛ/; needs a dictionary
+é → e ;
+è → ɛ ;
+f → f ;
+ge → ɟ ;
+gh → ɡ ;
+gi → ɟ ;
+{gl} [aeou] → ɡl ;
+{gl} → ʎ ;
+gn → ɲ ;
+g → ɡ ;
+h → ;
+iau → ɪa̯ʊ̯ ;
+ia → ɪa̯ ;
+ie → ɪɛ̯ ;
+iu → ɪʊ̯ ;
+i → i ;
+j → j ;
+k → k ;
+l → l ;
+m → m ;
+n → n ;
+o → ɔ ;
+p → p ;
+q → k ;
+r → r ;
+{sch} [aeiou] → ʒ ;  # can also be /ʃ/; needs a dictionary
+{sch} → ʃ ;  # can also be /ʒ/; needs a dictionary
+{s} [cptnm] → ʃ ;
+{s} [gbdv] → ʒ ;
+s → s ;  # can also be /z/; needs a dictionary
+tg → c ;
+tsch → t͡ʃ ;
+t → t ;
+uau → ʊa̯ʊ̯ ;
+ua → ʊa̯ ;
+uei → ʊɛ̯ɪ̯ ;
+ue → ʊɛ̯ ;
+uo → ʊɔ̯ ;
+u → u ;
+v → v ;
+w → v ;
+x → ks ;
+y → i ;
+z → t͡s ;
+::NULL;
+mm+ → mː;
+nn+ → nː;
+ɲɲ+ → ɲː;
+pp+ → pː;
+bb+ → bː;
+tt+ → tː;
+dd+ → dː;
+cc+ → cː ;
+ɟɟ+ → ɟː ;
+kk+ → kː ;
+ɡɡ+ → ɡː ;
+ff+ → fː ;
+vv+ → vː ;
+ss+ → sː ;
+zz+ → zː ;
+ʃʃ+ → ʃː ;
+ʒʒ+ → ʒː ;
+rr+ → rː ;
+ll+ → lː ;
+jj+ → jː ;
+

package/transforms/ro-am.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "am-t-ro",
+  "_source": "ro",
+  "_direction": "forward",
+  "_target": "am",
+  "_rulesFile": "ro-am.txt"
+}

package/transforms/ro-am.txt

@@ -0,0 +1,4 @@
+
+::ro-ro_FONIPA;
+::am_FONIPA-am;
+

package/transforms/ro-ar.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "ar-t-ro",
+  "_source": "ro",
+  "_direction": "forward",
+  "_target": "ar",
+  "_rulesFile": "ro-ar.txt"
+}

package/transforms/ro-ar.txt

@@ -0,0 +1,4 @@
+
+::ro-ro_FONIPA;
+::und_FONIPA-ar;
+

package/transforms/ro-chr.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "chr-t-ro",
+  "_source": "ro",
+  "_direction": "forward",
+  "_target": "chr",
+  "_rulesFile": "ro-chr.txt"
+}

package/transforms/ro-chr.txt

@@ -0,0 +1,4 @@
+
+::ro-ro_FONIPA;
+::und_FONIPA-chr;
+

package/transforms/ro-fa.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "fa-t-ro",
+  "_source": "ro",
+  "_direction": "forward",
+  "_target": "fa",
+  "_rulesFile": "ro-fa.txt"
+}

package/transforms/ro-fa.txt

@@ -0,0 +1,4 @@
+
+::ro-ro_FONIPA;
+::und_FONIPA-fa;
+

package/transforms/ro-ja.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "ja-t-ro",
+  "_source": "ro",
+  "_direction": "forward",
+  "_target": "ja",
+  "_rulesFile": "ro-ja.txt"
+}

package/transforms/ro-ja.txt

@@ -0,0 +1,4 @@
+
+::ro-ro_FONIPA;
+::ro_FONIPA-ja;
+

package/transforms/ro-ro_FONIPA.json

@@ -0,0 +1,8 @@
+{
+  "_visibility": "external",
+  "_aliasBcp47": "ro-fonipa-t-ro",
+  "_source": "ro",
+  "_direction": "forward",
+  "_target": "ro_FONIPA",
+  "_rulesFile": "ro-ro_FONIPA.txt"
+}