#REFACTORING by exept
class: Smalltalk class
changed: #recursiveInstallAutoloadedClassesFrom:rememberIn:maxLevels:noAutoload:packageTop:showSplashInLevels:
Transcript showCR:(... bindWith:...) -> Transcript showCR:... with:...
"
COPYRIGHT (c) 2004 by eXept Software AG
All Rights Reserved
This software is furnished under a license and may be used
only in accordance with the terms of that license and with the
inclusion of the above copyright notice. This software may not
be provided or otherwise made available to, or used by, any
other person. No title to or ownership of the software is
hereby transferred.
"
"{ Package: 'stx:libbasic' }"
"{ NameSpace: CharacterEncoderImplementations }"
KOI8_R subclass:#KOI8_U
instanceVariableNames:''
classVariableNames:''
poolDictionaries:''
category:'Collections-Text-Encodings'
!
!KOI8_U class methodsFor:'documentation'!
copyright
"
COPYRIGHT (c) 2004 by eXept Software AG
All Rights Reserved
This software is furnished under a license and may be used
only in accordance with the terms of that license and with the
inclusion of the above copyright notice. This software may not
be provided or otherwise made available to, or used by, any
other person. No title to or ownership of the software is
hereby transferred.
"
!
documentation
"
I encode/decode KOI8-U encoded documents (cyrillic - ukraine)
"
! !
!KOI8_U class methodsFor:'mapping'!
mapFileURL2_relativePathName
^ 'KOI8-U'
!
mapping
"
# From: http://std.dkuug.dk/i18n/charmaps/KOI8-U
<code_set_name> KOI8-U
<comment_char> %
<escape_char> /
% version: 1.0
% repertoiremap: mnemonic,ds
% source: RFC 2319
% Mibenum: 2088
% source: http://www.net.ua/KOI8-U/
CHARMAP
<NU> /x00 <U0000> NULL (NUL)
<SH> /x01 <U0001> START OF HEADING (SOH)
<SX> /x02 <U0002> START OF TEXT (STX)
<EX> /x03 <U0003> END OF TEXT (ETX)
<ET> /x04 <U0004> END OF TRANSMISSION (EOT)
<EQ> /x05 <U0005> ENQUIRY (ENQ)
<AK> /x06 <U0006> ACKNOWLEDGE (ACK)
<BL> /x07 <U0007> BELL (BEL)
<BS> /x08 <U0008> BACKSPACE (BS)
<HT> /x09 <U0009> CHARACTER TABULATION (HT)
<LF> /x0A <U000A> LINE FEED (LF)
<VT> /x0B <U000B> LINE TABULATION (VT)
<FF> /x0C <U000C> FORM FEED (FF)
<CR> /x0D <U000D> CARRIAGE RETURN (CR)
<SO> /x0E <U000E> SHIFT OUT (SO)
<SI> /x0F <U000F> SHIFT IN (SI)
<DL> /x10 <U0010> DATALINK ESCAPE (DLE)
<D1> /x11 <U0011> DEVICE CONTROL ONE (DC1)
<D2> /x12 <U0012> DEVICE CONTROL TWO (DC2)
<D3> /x13 <U0013> DEVICE CONTROL THREE (DC3)
<D4> /x14 <U0014> DEVICE CONTROL FOUR (DC4)
<NK> /x15 <U0015> NEGATIVE ACKNOWLEDGE (NAK)
<SY> /x16 <U0016> SYNCHRONOUS IDLE (SYN)
<EB> /x17 <U0017> END OF TRANSMISSION BLOCK (ETB)
<CN> /x18 <U0018> CANCEL (CAN)
<EM> /x19 <U0019> END OF MEDIUM (EM)
<SB> /x1A <U001A> SUBSTITUTE (SUB)
<EC> /x1B <U001B> ESCAPE (ESC)
<FS> /x1C <U001C> FILE SEPARATOR (IS4)
<GS> /x1D <U001D> GROUP SEPARATOR (IS3)
<RS> /x1E <U001E> RECORD SEPARATOR (IS2)
<US> /x1F <U001F> UNIT SEPARATOR (IS1)
<SP> /x20 <U0020> SPACE
<!!> /x21 <U0021> EXCLAMATION MARK
<'> /x22 <U0022> QUOTATION MARK
<Nb> /x23 <U0023> NUMBER SIGN
<DO> /x24 <U0024> DOLLAR SIGN
<%> /x25 <U0025> PERCENT SIGN
<&> /x26 <U0026> AMPERSAND
<'> /x27 <U0027> APOSTROPHE
<(> /x28 <U0028> LEFT PARENTHESIS
<)> /x29 <U0029> RIGHT PARENTHESIS
<*> /x2A <U002A> ASTERISK
<+> /x2B <U002B> PLUS SIGN
<,> /x2C <U002C> COMMA
<-> /x2D <U002D> HYPHEN-MINUS
<.> /x2E <U002E> FULL STOP
<//> /x2F <U002F> SOLIDUS
<0> /x30 <U0030> DIGIT ZERO
<1> /x31 <U0031> DIGIT ONE
<2> /x32 <U0032> DIGIT TWO
<3> /x33 <U0033> DIGIT THREE
<4> /x34 <U0034> DIGIT FOUR
<5> /x35 <U0035> DIGIT FIVE
<6> /x36 <U0036> DIGIT SIX
<7> /x37 <U0037> DIGIT SEVEN
<8> /x38 <U0038> DIGIT EIGHT
<9> /x39 <U0039> DIGIT NINE
<:> /x3A <U003A> COLON
<;> /x3B <U003B> SEMICOLON
<<> /x3C <U003C> LESS-THAN SIGN
<=> /x3D <U003D> EQUALS SIGN
</>> /x3E <U003E> GREATER-THAN SIGN
<?> /x3F <U003F> QUESTION MARK
<At> /x40 <U0040> COMMERCIAL AT
<A> /x41 <U0041> LATIN CAPITAL LETTER A
<B> /x42 <U0042> LATIN CAPITAL LETTER B
<C> /x43 <U0043> LATIN CAPITAL LETTER C
<D> /x44 <U0044> LATIN CAPITAL LETTER D
<E> /x45 <U0045> LATIN CAPITAL LETTER E
<F> /x46 <U0046> LATIN CAPITAL LETTER F
<G> /x47 <U0047> LATIN CAPITAL LETTER G
<H> /x48 <U0048> LATIN CAPITAL LETTER H
<I> /x49 <U0049> LATIN CAPITAL LETTER I
<J> /x4A <U004A> LATIN CAPITAL LETTER J
<K> /x4B <U004B> LATIN CAPITAL LETTER K
<L> /x4C <U004C> LATIN CAPITAL LETTER L
<M> /x4D <U004D> LATIN CAPITAL LETTER M
<N> /x4E <U004E> LATIN CAPITAL LETTER N
<O> /x4F <U004F> LATIN CAPITAL LETTER O
<P> /x50 <U0050> LATIN CAPITAL LETTER P
<Q> /x51 <U0051> LATIN CAPITAL LETTER Q
<R> /x52 <U0052> LATIN CAPITAL LETTER R
<S> /x53 <U0053> LATIN CAPITAL LETTER S
<T> /x54 <U0054> LATIN CAPITAL LETTER T
<U> /x55 <U0055> LATIN CAPITAL LETTER U
<V> /x56 <U0056> LATIN CAPITAL LETTER V
<W> /x57 <U0057> LATIN CAPITAL LETTER W
<X> /x58 <U0058> LATIN CAPITAL LETTER X
<Y> /x59 <U0059> LATIN CAPITAL LETTER Y
<Z> /x5A <U005A> LATIN CAPITAL LETTER Z
<<(> /x5B <U005B> LEFT SQUARE BRACKET
<////> /x5C <U005C> REVERSE SOLIDUS
<)/>> /x5D <U005D> RIGHT SQUARE BRACKET
<'/>> /x5E <U005E> CIRCUMFLEX ACCENT
<_> /x5F <U005F> LOW LINE
<'!!> /x60 <U0060> GRAVE ACCENT
<a> /x61 <U0061> LATIN SMALL LETTER A
<b> /x62 <U0062> LATIN SMALL LETTER B
<c> /x63 <U0063> LATIN SMALL LETTER C
<d> /x64 <U0064> LATIN SMALL LETTER D
<e> /x65 <U0065> LATIN SMALL LETTER E
<f> /x66 <U0066> LATIN SMALL LETTER F
<g> /x67 <U0067> LATIN SMALL LETTER G
<h> /x68 <U0068> LATIN SMALL LETTER H
<i> /x69 <U0069> LATIN SMALL LETTER I
<j> /x6A <U006A> LATIN SMALL LETTER J
<k> /x6B <U006B> LATIN SMALL LETTER K
<l> /x6C <U006C> LATIN SMALL LETTER L
<m> /x6D <U006D> LATIN SMALL LETTER M
<n> /x6E <U006E> LATIN SMALL LETTER N
<o> /x6F <U006F> LATIN SMALL LETTER O
<p> /x70 <U0070> LATIN SMALL LETTER P
<q> /x71 <U0071> LATIN SMALL LETTER Q
<r> /x72 <U0072> LATIN SMALL LETTER R
<s> /x73 <U0073> LATIN SMALL LETTER S
<t> /x74 <U0074> LATIN SMALL LETTER T
<u> /x75 <U0075> LATIN SMALL LETTER U
<v> /x76 <U0076> LATIN SMALL LETTER V
<w> /x77 <U0077> LATIN SMALL LETTER W
<x> /x78 <U0078> LATIN SMALL LETTER X
<y> /x79 <U0079> LATIN SMALL LETTER Y
<z> /x7A <U007A> LATIN SMALL LETTER Z
<(!!> /x7B <U007B> LEFT CURLY BRACKET
<!!!!> /x7C <U007C> VERTICAL LINE
<!!)> /x7D <U007D> RIGHT CURLY BRACKET
<'?> /x7E <U007E> TILDE
<DT> /x7F <U007F> DELETE (DEL)
<hh> /x80 <U2500> BOX DRAWINGS LIGHT HORIZONTAL
<vv> /x81 <U2502> BOX DRAWINGS LIGHT VERTICAL
<dr> /x82 <U250C> BOX DRAWINGS LIGHT DOWN AND RIGHT
<dl> /x83 <U2510> BOX DRAWINGS LIGHT DOWN AND LEFT
<ur> /x84 <U2514> BOX DRAWINGS LIGHT UP AND RIGHT
<ul> /x85 <U2518> BOX DRAWINGS LIGHT UP AND LEFT
<vr> /x86 <U251C> BOX DRAWINGS LIGHT VERTICAL AND RIGHT
<vl> /x87 <U2524> BOX DRAWINGS LIGHT VERTICAL AND LEFT
<dh> /x88 <U252C> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL
<uh> /x89 <U2534> BOX DRAWINGS LIGHT UP AND HORIZONTAL
<vh> /x8A <U253C> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL
<TB> /x8B <U2580> UPPER HALF BLOCK
<LB> /x8C <U2584> LOWER HALF BLOCK
<FB> /x8D <U2588> FULL BLOCK
<lB> /x8E <U258C> LEFT HALF BLOCK
<RB> /x8F <U2590> RIGHT HALF BLOCK
<.S> /x90 <U2591> LIGHT SHADE
<:S> /x91 <U2592> MEDIUM SHADE
<?S> /x92 <U2593> DARK SHADE
<Iu> /x93 <U2320> TOP HALF INTEGRAL
<fS> /x94 <U25A0> BLACK SQUARE
<Sb> /x95 <U2219> BULLET OPERATOR
<RT> /x96 <U221A> SQUARE ROOT
<?2> /x97 <U2248> ALMOST EQUAL TO
<=<> /x98 <U2264> LESS-THAN OR EQUAL TO
</>=> /x99 <U2265> GREATER-THAN OR EQUAL TO
<NS> /x9A <U00A0> NO-BREAK SPACE
<Il> /x9B <U2321> BOTTOM HALF INTEGRAL
<DG> /x9C <U00B0> DEGREE SIGN
<2S> /x9D <U00B2> SUPERSCRIPT TWO
<.M> /x9E <U00B7> MIDDLE DOT
<-:> /x9F <U00F7> DIVISION SIGN
<HH> /xA0 <U2550> BOX DRAWINGS DOUBLE HORIZONTAL
<VV> /xA1 <U2551> BOX DRAWINGS DOUBLE VERTICAL
<dR> /xA2 <U2552> BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE
<io> /xA3 <U0451> CYRILLIC SMALL LETTER IO
<ie> /xA4 <U0454> CYRILLIC SMALL LETTER UKRAINIAN IE
<DR> /xA5 <U2554> BOX DRAWINGS DOUBLE DOWN AND RIGHT
<ii> /xA6 <U0456> CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
<yi> /xA7 <U0457> CYRILLIC SMALL LETTER YI (Ukrainian)
<LD> /xA8 <U2557> BOX DRAWINGS DOUBLE DOWN AND LEFT
<uR> /xA9 <U2558> BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE
<Ur> /xAA <U2559> BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE
<UR> /xAB <U255A> BOX DRAWINGS DOUBLE UP AND RIGHT
<uL> /xAC <U255B> BOX DRAWINGS UP SINGLE AND LEFT DOUBLE
<g3> /xAD <U0491> CYRILLIC SMALL LETTER GHE WITH UPTURN
<UL> /xAE <U255D> BOX DRAWINGS DOUBLE UP AND LEFT
<vR> /xAF <U255E> BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE
<Vr> /xB0 <U255F> BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE
<VR> /xB1 <U2560> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT
<vL> /xB2 <U2561> BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE
<IO> /xB3 <U0401> CYRILLIC CAPITAL LETTER IO
<IE> /xB4 <U0404> CYRILLIC CAPITAL LETTER UKRAINIAN IE
<VL> /xB5 <U2563> BOX DRAWINGS DOUBLE VERTICAL AND LEFT
<II> /xB6 <U0406> CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
<YI> /xB7 <U0407> CYRILLIC CAPITAL LETTER YI (Ukrainian)
<DH> /xB8 <U2566> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL
<uH> /xB9 <U2567> BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE
<Uh> /xBA <U2568> BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE
<UH> /xBB <U2569> BOX DRAWINGS DOUBLE UP AND HORIZONTAL
<vH> /xBC <U256A> BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE
<G3> /xBD <U0490> CYRILLIC CAPITAL LETTER GHE WITH UPTURN
<VH> /xBE <U256C> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL
<Co> /xBF <U00A9> COPYRIGHT SIGN
<ju> /xC0 <U044E> CYRILLIC SMALL LETTER YU
<a=> /xC1 <U0430> CYRILLIC SMALL LETTER A
<b=> /xC2 <U0431> CYRILLIC SMALL LETTER BE
<c=> /xC3 <U0446> CYRILLIC SMALL LETTER TSE
<d=> /xC4 <U0434> CYRILLIC SMALL LETTER DE
<e=> /xC5 <U0435> CYRILLIC SMALL LETTER IE
<f=> /xC6 <U0444> CYRILLIC SMALL LETTER EF
<g=> /xC7 <U0433> CYRILLIC SMALL LETTER GHE
<h=> /xC8 <U0445> CYRILLIC SMALL LETTER HA
<i=> /xC9 <U0438> CYRILLIC SMALL LETTER I
<j=> /xCA <U0439> CYRILLIC SMALL LETTER SHORT I
<k=> /xCB <U043A> CYRILLIC SMALL LETTER KA
<l=> /xCC <U043B> CYRILLIC SMALL LETTER EL
<m=> /xCD <U043C> CYRILLIC SMALL LETTER EM
<n=> /xCE <U043D> CYRILLIC SMALL LETTER EN
<o=> /xCF <U043E> CYRILLIC SMALL LETTER O
<p=> /xD0 <U043F> CYRILLIC SMALL LETTER PE
<ja> /xD1 <U044F> CYRILLIC SMALL LETTER YA
<r=> /xD2 <U0440> CYRILLIC SMALL LETTER ER
<s=> /xD3 <U0441> CYRILLIC SMALL LETTER ES
<t=> /xD4 <U0442> CYRILLIC SMALL LETTER TE
<u=> /xD5 <U0443> CYRILLIC SMALL LETTER U
<z%> /xD6 <U0436> CYRILLIC SMALL LETTER ZHE
<v=> /xD7 <U0432> CYRILLIC SMALL LETTER VE
<%'> /xD8 <U044C> CYRILLIC SMALL LETTER SOFT SIGN
<y=> /xD9 <U044B> CYRILLIC SMALL LETTER YERU
<z=> /xDA <U0437> CYRILLIC SMALL LETTER ZE
<s%> /xDB <U0448> CYRILLIC SMALL LETTER SHA
<je> /xDC <U044D> CYRILLIC SMALL LETTER E
<sc> /xDD <U0449> CYRILLIC SMALL LETTER SHCHA
<c%> /xDE <U0447> CYRILLIC SMALL LETTER CHE
<='> /xDF <U044A> CYRILLIC SMALL LETTER HARD SIGN
<JU> /xE0 <U042E> CYRILLIC CAPITAL LETTER YU
<A=> /xE1 <U0410> CYRILLIC CAPITAL LETTER A
<B=> /xE2 <U0411> CYRILLIC CAPITAL LETTER BE
<C=> /xE3 <U0426> CYRILLIC CAPITAL LETTER TSE
<D=> /xE4 <U0414> CYRILLIC CAPITAL LETTER DE
<E=> /xE5 <U0415> CYRILLIC CAPITAL LETTER IE
<F=> /xE6 <U0424> CYRILLIC CAPITAL LETTER EF
<G=> /xE7 <U0413> CYRILLIC CAPITAL LETTER GHE
<H=> /xE8 <U0425> CYRILLIC CAPITAL LETTER HA
<I=> /xE9 <U0418> CYRILLIC CAPITAL LETTER I
<J=> /xEA <U0419> CYRILLIC CAPITAL LETTER SHORT I
<K=> /xEB <U041A> CYRILLIC CAPITAL LETTER KA
<L=> /xEC <U041B> CYRILLIC CAPITAL LETTER EL
<M=> /xED <U041C> CYRILLIC CAPITAL LETTER EM
<N=> /xEE <U041D> CYRILLIC CAPITAL LETTER EN
<O=> /xEF <U041E> CYRILLIC CAPITAL LETTER O
<P=> /xF0 <U041F> CYRILLIC CAPITAL LETTER PE
<JA> /xF1 <U042F> CYRILLIC CAPITAL LETTER YA
<R=> /xF2 <U0420> CYRILLIC CAPITAL LETTER ER
<S=> /xF3 <U0421> CYRILLIC CAPITAL LETTER ES
<T=> /xF4 <U0422> CYRILLIC CAPITAL LETTER TE
<U=> /xF5 <U0423> CYRILLIC CAPITAL LETTER U
<Z%> /xF6 <U0416> CYRILLIC CAPITAL LETTER ZHE
<V=> /xF7 <U0412> CYRILLIC CAPITAL LETTER VE
<%'> /xF8 <U042C> CYRILLIC CAPITAL LETTER SOFT SIGN
<Y=> /xF9 <U042B> CYRILLIC CAPITAL LETTER YERU
<Z=> /xFA <U0417> CYRILLIC CAPITAL LETTER ZE
<S%> /xFB <U0428> CYRILLIC CAPITAL LETTER SHA
<JE> /xFC <U042D> CYRILLIC CAPITAL LETTER E
<Sc> /xFD <U0429> CYRILLIC CAPITAL LETTER SHCHA
<C%> /xFE <U0427> CYRILLIC CAPITAL LETTER CHE
<='> /xFF <U042A> CYRILLIC CAPITAL LETTER HARD SIGN
END CHARMAP
"
! !
!KOI8_U class methodsFor:'queries'!
userFriendlyNameOfEncoding
^ 'Cyrillic (Ukraine)'
! !
!KOI8_U methodsFor:'encoding & decoding'!
decode:codeArg
"Automagically generated by generateCode - do not modify.
Decode from my encoding into unicode."
|code "{ Class: SmallInteger }"|
code := codeArg.
code > 16rA3 ifTrue:[
code <= 16rBD ifTrue:[
code == 16rA4 ifTrue:[
^ "16r00A4" 16r0454 " CYRILLIC SMALL LETTER UKRAINIAN IE "
].
code == 16rA6 ifTrue:[
^ "16r00A6" 16r0456 " CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I "
].
code == 16rA7 ifTrue:[
^ "16r00A7" 16r0457 " CYRILLIC SMALL LETTER YI (Ukrainian) "
].
code == 16rAD ifTrue:[
^ "16r00AD" 16r0491 " CYRILLIC SMALL LETTER GHE WITH UPTURN "
].
code == 16rB4 ifTrue:[
^ "16r00B4" 16r0404 " CYRILLIC CAPITAL LETTER UKRAINIAN IE "
].
code == 16rB6 ifTrue:[
^ "16r00B6" 16r0406 " CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I "
].
code == 16rB7 ifTrue:[
^ "16r00B7" 16r0407 " CYRILLIC CAPITAL LETTER YI (Ukrainian) "
].
code == 16rBD ifTrue:[
^ "16r00BD" 16r0490 " CYRILLIC CAPITAL LETTER GHE WITH UPTURN "
].
].
].
^ super decode:code
!
encode:unicodeArg
"Automagically generated by generateCode - do not modify.
Encode from unicode into my encoding."
|unicode "{ Class: SmallInteger }"|
unicode := unicodeArg.
unicode > 16r403 ifTrue:[
unicode <= 16r256B ifTrue:[
((unicode == 16r2553)
or:[(unicode == 16r255C)
or:[(unicode == 16r2564)
or:[(unicode == 16r2565)
or:[(unicode == 16r2555)
or:[(unicode == 16r2556)
or:[(unicode == 16r2562)
or:[(unicode == 16r256B)]]]]]]]) ifTrue:[
^ self decodingError.
].
unicode == 16r404 ifTrue:[
^ "16r0404" 16r00B4 " CYRILLIC CAPITAL LETTER UKRAINIAN IE "
].
unicode == 16r406 ifTrue:[
^ "16r0406" 16r00B6 " CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I "
].
unicode == 16r407 ifTrue:[
^ "16r0407" 16r00B7 " CYRILLIC CAPITAL LETTER YI (Ukrainian) "
].
unicode == 16r454 ifTrue:[
^ "16r0454" 16r00A4 " CYRILLIC SMALL LETTER UKRAINIAN IE "
].
unicode == 16r456 ifTrue:[
^ "16r0456" 16r00A6 " CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I "
].
unicode == 16r457 ifTrue:[
^ "16r0457" 16r00A7 " CYRILLIC SMALL LETTER YI (Ukrainian) "
].
unicode == 16r490 ifTrue:[
^ "16r0490" 16r00BD " CYRILLIC CAPITAL LETTER GHE WITH UPTURN "
].
unicode == 16r491 ifTrue:[
^ "16r0491" 16r00AD " CYRILLIC SMALL LETTER GHE WITH UPTURN "
].
].
].
^ super encode:unicode
! !
!KOI8_U class methodsFor:'documentation'!
version
^ '$Header$'
! !