diff options
Diffstat (limited to 'app/nginx/contrib/unicode2nginx')
-rw-r--r-- | app/nginx/contrib/unicode2nginx/koi-utf | 131 | ||||
-rwxr-xr-x | app/nginx/contrib/unicode2nginx/unicode-to-nginx.pl | 48 | ||||
-rw-r--r-- | app/nginx/contrib/unicode2nginx/win-utf | 130 |
3 files changed, 309 insertions, 0 deletions
diff --git a/app/nginx/contrib/unicode2nginx/koi-utf b/app/nginx/contrib/unicode2nginx/koi-utf new file mode 100644 index 0000000..48853af --- /dev/null +++ b/app/nginx/contrib/unicode2nginx/koi-utf @@ -0,0 +1,131 @@ +charset_map koi8-r utf-8 { + + 80 E29480 ; # BOX DRAWINGS LIGHT HORIZONTAL + 81 E29482 ; # BOX DRAWINGS LIGHT VERTICAL + 82 E2948C ; # BOX DRAWINGS LIGHT DOWN AND RIGHT + 83 E29490 ; # BOX DRAWINGS LIGHT DOWN AND LEFT + 84 E29494 ; # BOX DRAWINGS LIGHT UP AND RIGHT + 85 E29498 ; # BOX DRAWINGS LIGHT UP AND LEFT + 86 E2949C ; # BOX DRAWINGS LIGHT VERTICAL AND RIGHT + 87 E294A4 ; # BOX DRAWINGS LIGHT VERTICAL AND LEFT + 88 E294AC ; # BOX DRAWINGS LIGHT DOWN AND HORIZONTAL + 89 E294B4 ; # BOX DRAWINGS LIGHT UP AND HORIZONTAL + 8A E294BC ; # BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL + 8B E29680 ; # UPPER HALF BLOCK + 8C E29684 ; # LOWER HALF BLOCK + 8D E29688 ; # FULL BLOCK + 8E E2968C ; # LEFT HALF BLOCK + 8F E29690 ; # RIGHT HALF BLOCK + 90 E29691 ; # LIGHT SHADE + 91 E29692 ; # MEDIUM SHADE + 92 E29693 ; # DARK SHADE + 93 E28CA0 ; # TOP HALF INTEGRAL + 94 E296A0 ; # BLACK SQUARE + 95 E28899 ; # BULLET OPERATOR + 96 E2889A ; # SQUARE ROOT + 97 E28988 ; # ALMOST EQUAL TO + 98 E289A4 ; # LESS-THAN OR EQUAL TO + 99 E289A5 ; # GREATER-THAN OR EQUAL TO + 9A C2A0 ; # NO-BREAK SPACE + 9B E28CA1 ; # BOTTOM HALF INTEGRAL + 9C C2B0 ; # DEGREE SIGN + 9D C2B2 ; # SUPERSCRIPT TWO + 9E C2B7 ; # MIDDLE DOT + 9F C3B7 ; # DIVISION SIGN + A0 E29590 ; # BOX DRAWINGS DOUBLE HORIZONTAL + A1 E29591 ; # BOX DRAWINGS DOUBLE VERTICAL + A2 E29592 ; # BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE + A3 D191 ; # CYRILLIC SMALL LETTER IO + A4 E29593 ; # BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE + A5 E29594 ; # BOX DRAWINGS DOUBLE DOWN AND RIGHT + A6 E29595 ; # BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE + A7 E29596 ; # BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE + A8 E29597 ; # BOX DRAWINGS DOUBLE DOWN AND LEFT + A9 E29598 ; # BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE + AA E29599 ; # BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE + AB E2959A ; # BOX DRAWINGS DOUBLE UP AND RIGHT + AC E2959B ; # BOX DRAWINGS UP SINGLE AND LEFT DOUBLE + AD E2959C ; # BOX DRAWINGS UP DOUBLE AND LEFT SINGLE + AE E2959D ; # BOX DRAWINGS DOUBLE UP AND LEFT + AF E2959E ; # BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE + B0 E2959F ; # BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE + B1 E295A0 ; # BOX DRAWINGS DOUBLE VERTICAL AND RIGHT + B2 E295A1 ; # BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE + B3 D081 ; # CYRILLIC CAPITAL LETTER IO + B4 E295A2 ; # BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE + B5 E295A3 ; # BOX DRAWINGS DOUBLE VERTICAL AND LEFT + B6 E295A4 ; # BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE + B7 E295A5 ; # BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE + B8 E295A6 ; # BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL + B9 E295A7 ; # BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE + BA E295A8 ; # BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE + BB E295A9 ; # BOX DRAWINGS DOUBLE UP AND HORIZONTAL + BC E295AA ; # BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE + BD E295AB ; # BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE + BE E295AC ; # BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL + BF C2A9 ; # COPYRIGHT SIGN + C0 D18E ; # CYRILLIC SMALL LETTER YU + C1 D0B0 ; # CYRILLIC SMALL LETTER A + C2 D0B1 ; # CYRILLIC SMALL LETTER BE + C3 D186 ; # CYRILLIC SMALL LETTER TSE + C4 D0B4 ; # CYRILLIC SMALL LETTER DE + C5 D0B5 ; # CYRILLIC SMALL LETTER IE + C6 D184 ; # CYRILLIC SMALL LETTER EF + C7 D0B3 ; # CYRILLIC SMALL LETTER GHE + C8 D185 ; # CYRILLIC SMALL LETTER HA + C9 D0B8 ; # CYRILLIC SMALL LETTER I + CA D0B9 ; # CYRILLIC SMALL LETTER SHORT I + CB D0BA ; # CYRILLIC SMALL LETTER KA + CC D0BB ; # CYRILLIC SMALL LETTER EL + CD D0BC ; # CYRILLIC SMALL LETTER EM + CE D0BD ; # CYRILLIC SMALL LETTER EN + CF D0BE ; # CYRILLIC SMALL LETTER O + D0 D0BF ; # CYRILLIC SMALL LETTER PE + D1 D18F ; # CYRILLIC SMALL LETTER YA + D2 D180 ; # CYRILLIC SMALL LETTER ER + D3 D181 ; # CYRILLIC SMALL LETTER ES + D4 D182 ; # CYRILLIC SMALL LETTER TE + D5 D183 ; # CYRILLIC SMALL LETTER U + D6 D0B6 ; # CYRILLIC SMALL LETTER ZHE + D7 D0B2 ; # CYRILLIC SMALL LETTER VE + D8 D18C ; # CYRILLIC SMALL LETTER SOFT SIGN + D9 D18B ; # CYRILLIC SMALL LETTER YERU + DA D0B7 ; # CYRILLIC SMALL LETTER ZE + DB D188 ; # CYRILLIC SMALL LETTER SHA + DC D18D ; # CYRILLIC SMALL LETTER E + DD D189 ; # CYRILLIC SMALL LETTER SHCHA + DE D187 ; # CYRILLIC SMALL LETTER CHE + DF D18A ; # CYRILLIC SMALL LETTER HARD SIGN + E0 D0AE ; # CYRILLIC CAPITAL LETTER YU + E1 D090 ; # CYRILLIC CAPITAL LETTER A + E2 D091 ; # CYRILLIC CAPITAL LETTER BE + E3 D0A6 ; # CYRILLIC CAPITAL LETTER TSE + E4 D094 ; # CYRILLIC CAPITAL LETTER DE + E5 D095 ; # CYRILLIC CAPITAL LETTER IE + E6 D0A4 ; # CYRILLIC CAPITAL LETTER EF + E7 D093 ; # CYRILLIC CAPITAL LETTER GHE + E8 D0A5 ; # CYRILLIC CAPITAL LETTER HA + E9 D098 ; # CYRILLIC CAPITAL LETTER I + EA D099 ; # CYRILLIC CAPITAL LETTER SHORT I + EB D09A ; # CYRILLIC CAPITAL LETTER KA + EC D09B ; # CYRILLIC CAPITAL LETTER EL + ED D09C ; # CYRILLIC CAPITAL LETTER EM + EE D09D ; # CYRILLIC CAPITAL LETTER EN + EF D09E ; # CYRILLIC CAPITAL LETTER O + F0 D09F ; # CYRILLIC CAPITAL LETTER PE + F1 D0AF ; # CYRILLIC CAPITAL LETTER YA + F2 D0A0 ; # CYRILLIC CAPITAL LETTER ER + F3 D0A1 ; # CYRILLIC CAPITAL LETTER ES + F4 D0A2 ; # CYRILLIC CAPITAL LETTER TE + F5 D0A3 ; # CYRILLIC CAPITAL LETTER U + F6 D096 ; # CYRILLIC CAPITAL LETTER ZHE + F7 D092 ; # CYRILLIC CAPITAL LETTER VE + F8 D0AC ; # CYRILLIC CAPITAL LETTER SOFT SIGN + F9 D0AB ; # CYRILLIC CAPITAL LETTER YERU + FA D097 ; # CYRILLIC CAPITAL LETTER ZE + FB D0A8 ; # CYRILLIC CAPITAL LETTER SHA + FC D0AD ; # CYRILLIC CAPITAL LETTER E + FD D0A9 ; # CYRILLIC CAPITAL LETTER SHCHA + FE D0A7 ; # CYRILLIC CAPITAL LETTER CHE + FF D0AA ; # CYRILLIC CAPITAL LETTER HARD SIGN +} diff --git a/app/nginx/contrib/unicode2nginx/unicode-to-nginx.pl b/app/nginx/contrib/unicode2nginx/unicode-to-nginx.pl new file mode 100755 index 0000000..d113fed --- /dev/null +++ b/app/nginx/contrib/unicode2nginx/unicode-to-nginx.pl @@ -0,0 +1,48 @@ +#!/usr/bin/perl -w + +# Convert unicode mappings to nginx configuration file format. + +# You may find useful mappings in various places, including +# unicode.org official site: +# +# http://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1251.TXT +# http://www.unicode.org/Public/MAPPINGS/VENDORS/MISC/KOI8-R.TXT + +# Needs perl 5.6 or later. + +# Written by Maxim Dounin, mdounin@mdounin.ru + +############################################################################### + +require 5.006; + +while (<>) { + # Skip comments and empty lines + + next if /^#/; + next if /^\s*$/; + chomp; + + # Convert mappings + + if (/^\s*0x(..)\s*0x(....)\s*(#.*)/) { + # Mapping <from-code> <unicode-code> "#" <unicode-name> + my $cs_code = $1; + my $un_code = $2; + my $un_name = $3; + + # Produce UTF-8 sequence from character code; + + my $un_utf8 = join('', + map { sprintf("%02X", $_) } + unpack("U0C*", pack("U", hex($un_code))) + ); + + print " $cs_code $un_utf8 ; $un_name\n"; + + } else { + warn "Unrecognized line: '$_'"; + } +} + +############################################################################### diff --git a/app/nginx/contrib/unicode2nginx/win-utf b/app/nginx/contrib/unicode2nginx/win-utf new file mode 100644 index 0000000..af9f9aa --- /dev/null +++ b/app/nginx/contrib/unicode2nginx/win-utf @@ -0,0 +1,130 @@ +charset_map windows-1251 utf-8 { + + 80 D082 ; #CYRILLIC CAPITAL LETTER DJE + 81 D083 ; #CYRILLIC CAPITAL LETTER GJE + 82 E2809A ; #SINGLE LOW-9 QUOTATION MARK + 83 D193 ; #CYRILLIC SMALL LETTER GJE + 84 E2809E ; #DOUBLE LOW-9 QUOTATION MARK + 85 E280A6 ; #HORIZONTAL ELLIPSIS + 86 E280A0 ; #DAGGER + 87 E280A1 ; #DOUBLE DAGGER + 88 E282AC ; #EURO SIGN + 89 E280B0 ; #PER MILLE SIGN + 8A D089 ; #CYRILLIC CAPITAL LETTER LJE + 8B E280B9 ; #SINGLE LEFT-POINTING ANGLE QUOTATION MARK + 8C D08A ; #CYRILLIC CAPITAL LETTER NJE + 8D D08C ; #CYRILLIC CAPITAL LETTER KJE + 8E D08B ; #CYRILLIC CAPITAL LETTER TSHE + 8F D08F ; #CYRILLIC CAPITAL LETTER DZHE + 90 D192 ; #CYRILLIC SMALL LETTER DJE + 91 E28098 ; #LEFT SINGLE QUOTATION MARK + 92 E28099 ; #RIGHT SINGLE QUOTATION MARK + 93 E2809C ; #LEFT DOUBLE QUOTATION MARK + 94 E2809D ; #RIGHT DOUBLE QUOTATION MARK + 95 E280A2 ; #BULLET + 96 E28093 ; #EN DASH + 97 E28094 ; #EM DASH + 99 E284A2 ; #TRADE MARK SIGN + 9A D199 ; #CYRILLIC SMALL LETTER LJE + 9B E280BA ; #SINGLE RIGHT-POINTING ANGLE QUOTATION MARK + 9C D19A ; #CYRILLIC SMALL LETTER NJE + 9D D19C ; #CYRILLIC SMALL LETTER KJE + 9E D19B ; #CYRILLIC SMALL LETTER TSHE + 9F D19F ; #CYRILLIC SMALL LETTER DZHE + A0 C2A0 ; #NO-BREAK SPACE + A1 D08E ; #CYRILLIC CAPITAL LETTER SHORT U + A2 D19E ; #CYRILLIC SMALL LETTER SHORT U + A3 D088 ; #CYRILLIC CAPITAL LETTER JE + A4 C2A4 ; #CURRENCY SIGN + A5 D290 ; #CYRILLIC CAPITAL LETTER GHE WITH UPTURN + A6 C2A6 ; #BROKEN BAR + A7 C2A7 ; #SECTION SIGN + A8 D081 ; #CYRILLIC CAPITAL LETTER IO + A9 C2A9 ; #COPYRIGHT SIGN + AA D084 ; #CYRILLIC CAPITAL LETTER UKRAINIAN IE + AB C2AB ; #LEFT-POINTING DOUBLE ANGLE QUOTATION MARK + AC C2AC ; #NOT SIGN + AD C2AD ; #SOFT HYPHEN + AE C2AE ; #REGISTERED SIGN + AF D087 ; #CYRILLIC CAPITAL LETTER YI + B0 C2B0 ; #DEGREE SIGN + B1 C2B1 ; #PLUS-MINUS SIGN + B2 D086 ; #CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I + B3 D196 ; #CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I + B4 D291 ; #CYRILLIC SMALL LETTER GHE WITH UPTURN + B5 C2B5 ; #MICRO SIGN + B6 C2B6 ; #PILCROW SIGN + B7 C2B7 ; #MIDDLE DOT + B8 D191 ; #CYRILLIC SMALL LETTER IO + B9 E28496 ; #NUMERO SIGN + BA D194 ; #CYRILLIC SMALL LETTER UKRAINIAN IE + BB C2BB ; #RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK + BC D198 ; #CYRILLIC SMALL LETTER JE + BD D085 ; #CYRILLIC CAPITAL LETTER DZE + BE D195 ; #CYRILLIC SMALL LETTER DZE + BF D197 ; #CYRILLIC SMALL LETTER YI + C0 D090 ; #CYRILLIC CAPITAL LETTER A + C1 D091 ; #CYRILLIC CAPITAL LETTER BE + C2 D092 ; #CYRILLIC CAPITAL LETTER VE + C3 D093 ; #CYRILLIC CAPITAL LETTER GHE + C4 D094 ; #CYRILLIC CAPITAL LETTER DE + C5 D095 ; #CYRILLIC CAPITAL LETTER IE + C6 D096 ; #CYRILLIC CAPITAL LETTER ZHE + C7 D097 ; #CYRILLIC CAPITAL LETTER ZE + C8 D098 ; #CYRILLIC CAPITAL LETTER I + C9 D099 ; #CYRILLIC CAPITAL LETTER SHORT I + CA D09A ; #CYRILLIC CAPITAL LETTER KA + CB D09B ; #CYRILLIC CAPITAL LETTER EL + CC D09C ; #CYRILLIC CAPITAL LETTER EM + CD D09D ; #CYRILLIC CAPITAL LETTER EN + CE D09E ; #CYRILLIC CAPITAL LETTER O + CF D09F ; #CYRILLIC CAPITAL LETTER PE + D0 D0A0 ; #CYRILLIC CAPITAL LETTER ER + D1 D0A1 ; #CYRILLIC CAPITAL LETTER ES + D2 D0A2 ; #CYRILLIC CAPITAL LETTER TE + D3 D0A3 ; #CYRILLIC CAPITAL LETTER U + D4 D0A4 ; #CYRILLIC CAPITAL LETTER EF + D5 D0A5 ; #CYRILLIC CAPITAL LETTER HA + D6 D0A6 ; #CYRILLIC CAPITAL LETTER TSE + D7 D0A7 ; #CYRILLIC CAPITAL LETTER CHE + D8 D0A8 ; #CYRILLIC CAPITAL LETTER SHA + D9 D0A9 ; #CYRILLIC CAPITAL LETTER SHCHA + DA D0AA ; #CYRILLIC CAPITAL LETTER HARD SIGN + DB D0AB ; #CYRILLIC CAPITAL LETTER YERU + DC D0AC ; #CYRILLIC CAPITAL LETTER SOFT SIGN + DD D0AD ; #CYRILLIC CAPITAL LETTER E + DE D0AE ; #CYRILLIC CAPITAL LETTER YU + DF D0AF ; #CYRILLIC CAPITAL LETTER YA + E0 D0B0 ; #CYRILLIC SMALL LETTER A + E1 D0B1 ; #CYRILLIC SMALL LETTER BE + E2 D0B2 ; #CYRILLIC SMALL LETTER VE + E3 D0B3 ; #CYRILLIC SMALL LETTER GHE + E4 D0B4 ; #CYRILLIC SMALL LETTER DE + E5 D0B5 ; #CYRILLIC SMALL LETTER IE + E6 D0B6 ; #CYRILLIC SMALL LETTER ZHE + E7 D0B7 ; #CYRILLIC SMALL LETTER ZE + E8 D0B8 ; #CYRILLIC SMALL LETTER I + E9 D0B9 ; #CYRILLIC SMALL LETTER SHORT I + EA D0BA ; #CYRILLIC SMALL LETTER KA + EB D0BB ; #CYRILLIC SMALL LETTER EL + EC D0BC ; #CYRILLIC SMALL LETTER EM + ED D0BD ; #CYRILLIC SMALL LETTER EN + EE D0BE ; #CYRILLIC SMALL LETTER O + EF D0BF ; #CYRILLIC SMALL LETTER PE + F0 D180 ; #CYRILLIC SMALL LETTER ER + F1 D181 ; #CYRILLIC SMALL LETTER ES + F2 D182 ; #CYRILLIC SMALL LETTER TE + F3 D183 ; #CYRILLIC SMALL LETTER U + F4 D184 ; #CYRILLIC SMALL LETTER EF + F5 D185 ; #CYRILLIC SMALL LETTER HA + F6 D186 ; #CYRILLIC SMALL LETTER TSE + F7 D187 ; #CYRILLIC SMALL LETTER CHE + F8 D188 ; #CYRILLIC SMALL LETTER SHA + F9 D189 ; #CYRILLIC SMALL LETTER SHCHA + FA D18A ; #CYRILLIC SMALL LETTER HARD SIGN + FB D18B ; #CYRILLIC SMALL LETTER YERU + FC D18C ; #CYRILLIC SMALL LETTER SOFT SIGN + FD D18D ; #CYRILLIC SMALL LETTER E + FE D18E ; #CYRILLIC SMALL LETTER YU + FF D18F ; #CYRILLIC SMALL LETTER YA +} |