lexer grammar hyphenationpatterns;

fragment
NonPrintableCharacter
    :   '\u0000'..'\u001F'|'\u007F'
    ;

CommentedLine
    :   '#' ~NonPrintableCharacter* '\r'? '\n'
    ;

fragment
Hyphen
    :   '·'|'='|'@'
    ;

fragment
Delimiter
    :   ';'|'{'|'/'|'}'|'['|'|'|']'|'<'|'>'
    ;

fragment
Character
    :   ~(NonPrintableCharacter|Hyphen|Delimiter)
    ;

fragment
Characters
    :   (Character) Character+
    ;

fragment
Word
    :   (Character) (Character) Characters
    ;

fragment
HyphenationMark
    :   ('·'|'='+|'@'+)
    ;

fragment
HyphenationPoint
    :   ('<' Character* '|' Character+ '>')? HyphenationMark
    |   '{' Characters '/' Character+ HyphenationMark Character* '}'
    |   '[' ((Character+ HyphenationMark '|' HyphenationMark Character+) | (HyphenationMark Character+ '|' Character+ HyphenationMark)) ']'
    ;

fragment
ColumnNumber
    :   '2'..'9'
    ;

fragment
Pattern
    :   Word
    |   Characters (HyphenationPoint Character+)* HyphenationPoint Characters
    |   '-' ('-'|ColumnNumber) '-'
    ;

HyphenationPattern
    :   Word
        (';' Pattern)+
        '\r'? '\n'
    ;