blob: d049845859ca9838c9f5f210d9432a3daab1ab39 (
plain) (
tree)
|
|
// Any_Latin1
translit_Any_Latin1 {
Rule {
//NFKD then NFC; then NFD all non-Latin-1 codepoints
":: NFKC ();"
":: [^\u0000-\u00FF] NFD ();"
//remove non-spacing marks
"[:Mn:] > ;"
//change some non-Latin-1 codepoints to similar characters in Latin-1 range
" [\u2000-\u200A\u3000] > ' ' ;"
" [\u01C3\u2762]> '!' ;"
" \u203C > '!!' ;"
" [\u203D\u2048] > '?!' ;"
" [\u02BA\u030B\u030E\u2033\u3003\u201C-\u201F] > '\"' ;"
" [\u066A\u2030\u2031] > '%' ;"
" [\u02B9\u02BC\u02C8\u0301\u2032\u2018-\u201B] > '' ;"
" [\u066D\u2217\u2731] > '*' ;"
" [\u060C\u201A\u3001] > ',' ;"
" [\u2010-\u2013\u2212] > '-' ;"
" [\u2023\u06D4\u3002\u0589] > '.' ;"
" [\u0338\u2044\u2215] > '/' ;"
" \u2236 > ':' ;"
" \u061B > ';' ;"
" [\u2039\u2329\u3008] > '<' ;"
" \u2261 > '=' ;"
" [\u203A\u232A\u3009] > '>' ;"
" [\u037E\u061F] > '?' ;"
" \u2049 > '!?' ;"
" \u212C > B ;"
" [\u2102\u212D] > C ;"
" [\u2107\u2130] > E ;"
" [\u2131\u2132] > F ;"
" [\u210B\u210C\u210D] > H ;"
" [\u2110\u2111\u2160] > I ;"
" \u212A > K ;"
" \u2112 > L ;"
" \u2133 > L ;"
" \u2115 > N ;"
" \u2119 > P ;"
" \u211A > Q ;"
" [\u211B\u211C\u211D] > R ;"
" [\u2124\u2128] > Z ;"
" \u2216 > '\\' ;"
" [\u02C4\u02C6\u0302\u2303] > '^' ;"
" [\u02CD\u0331\u0332\u2017] > '_' ;"
" [\u02CB\u0300\u2035] > '`' ;"
" [\u212E\u212F] > e ;"
" [\u0261\u210A] > g ;"
" [\u04BB\u210E] > h ;"
" \u0131 > i ;"
" \u207F > n ;"
" \u2134 > o ;"
" \u01B6 > z ;"
" [\u01C0\u2223\u2758] > '|' ;"
" [\u02DC\u0303\u223C\uFF5E] > '~' ;"
" \u202F > \u00A0 ;"
" \uFFFD > \u001A;"
" \u20A4 > \u00A3 ;"
" [\u20A0-\u20AF] > \u00A4 ;"
" \u0308 > \u00A8 ;"
" \u2117 > \u00A9 ;"
" [\u226A\u300A] > \u00AB ;"
" \u2310 > \u00AC ;"
" \u1806 > \u00AD ;"
" [\u02C9\u0304\u0305] > \u00AF ;"
" [\u02DA\u030A\u2070\u2218] > \u00B0 ;"
" \u2213 > \u00B1 ;"
" [\u02B9\u02CA\u0301\u2032] > \u00B4 ;"
" [\u204B\u2761] > \u00B6 ;"
" [\u2022\u2024\u2027\u2219\u22C5\u30FB] > \u00B7 ;"
" \u0327 > \u00B8 ;"
" [\u226B\u300B] > \u00BB ;"
" \u2014 > '--';"
" \u2015 > '---';"
" \u2024 > '..';"
" \u2025 > '...';"
" \u2016 > '||';"
" \uFEFF > ;"
//change all spacing character to space
"[^\u0000-\u00FF] > ' ';"
}
}
|
