!636 !1000300 // class Lexicon main { psvm { String src = "class main { String s; }"; MyLexicon lex = new MyLexicon(); lex.lexicalRules(true); LineNumberReader source = new LineNumberReader(new StringReader(src)); for (Object a; (a = lex.grab(source)) != lex.$;) { System.out.println("grabbed: " + a); } } static class MyLexicon extends Lexicon { private static final String idChars1 = "_$?-#+"; private static final String idChars2 = "_$-.+"; void lexicalRules(boolean allowShortFloatingPoints) { // These rules are taken from Java10.java (part of gi-0.9 distribution) int INFINITY = -1; /** * 19.3 Terminals from section 3.6: White Space: [[:space:]] */ put("WHITE_SPACE", PosixClass.space()); /** * 19.3 Terminals from section 3.7: Comment */ put("COMMENT", new Union( // // Traditional Comment: /\*[^*]+(\*([^*/][^*]*)?)*\*/ // new Concatenation( new Singleton("/*"), new Concatenation( new Repetition(new NonMatch("*"), 1, INFINITY), new Concatenation( new Repetition( new Concatenation( new Singleton("*"), new Repetition(new Concatenation( new NonMatch("*/"), new Repetition(new NonMatch("*"), 0, INFINITY) ), 0, 1) ), 0, INFINITY ), new Singleton("*/") ))), new Union( /** * End Of Line Comment: //[^\n]*\n */ new Concatenation( new Singleton("//"), new Concatenation( new Repetition(new NonMatch("\n"), 0, INFINITY), new Singleton("\n") )), // // Documentation Comment: /\*\*(([^*/][^*]*)?\*)*/ // new Concatenation( new Singleton("/**"), new Concatenation( new Repetition( new Concatenation( new Repetition(new Concatenation( new NonMatch("*/"), new Repetition(new NonMatch("*"), 0, INFINITY) ), 0, 1), new Singleton("*") ), 0, INFINITY ), new Singleton("/") )) ))); put("IDENTIFIER", new Concatenation( new Union( PosixClass.alpha(), new Match(idChars1) ), new Repetition( new Union( PosixClass.alnum(), new Match(idChars2) ), 0, INFINITY ) )); /** * 19.3 Terminals from section 3.10.1: Integer Literal */ put("INTEGER_LITERAL", new Concatenation( new Union( /** * Decimal Integer Literal: 0|[1-9][[:digit:]]* */ new Singleton("0"), new Union( new Concatenation( new Range('1', '9'), new Repetition(PosixClass.digit(), 0, INFINITY) ), new Union( /** * Hexadecimal Integer Literal: 0[xX][[:xdigit:]]+ */ new Concatenation( new Singleton("0"), new Concatenation( new Match("xX"), new Repetition(PosixClass.xdigit(), 1, INFINITY) )), /** * Octal Integer Literal: 0[0-7]+ */ new Concatenation( new Singleton("0"), new Repetition(new Range('0', '7'), 1, INFINITY) ) ))), new Repetition(new Match("lL"), 0, 1) )); /** * 19.3 Terminals from Java section 3.10.5: String Literal - extended for multi-line strings */ put("STRING_LITERAL", new Concatenation( new Singleton("\""), new Concatenation( new Repetition( new Union( /** * Single Character: [^\r\n"\\] */ new NonMatch("\"\\"), // Java: new NonMatch("\r\n\"\\"), /** * Escape Sequence: \\([btnfr\"'\\]|[0-3]?[0-7]{1,2}) */ new Concatenation( new Singleton("\\"), new Union( new Match("btnfr\"'\\"), new Concatenation( new Repetition(new Range('0', '3'), 0, 1), new Repetition(new Range('0', '7'), 1, 2) ) ) ) ), 0, INFINITY ), new Singleton("\"") ))); /** * Addition for trees: Single-quote string literal * (is stored quoted with double-quotes. Yeah, it's * a little confusing, but don't worry.) */ put("SQ_STRING_LITERAL", new Concatenation( new Singleton("\'"), new Concatenation( new Repetition( new Union( /** * Single Character: [^\r\n"\\] */ new NonMatch("\r\n\'\\"), /** * Escape Sequence: \\([btnfr\"'\\]|[0-3]?[0-7]{1,2}) */ new Concatenation( new Singleton("\\"), new Union( new Match("btnfr\"'\\"), new Concatenation( new Repetition(new Range('0', '3'), 0, 1), new Repetition(new Range('0', '7'), 1, 2) ) ) ) ), 0, INFINITY ), new Singleton("\'") ))); /** * Multi-line string literals within brackets (>> hello world <<) */ put("ML_STRING_LITERAL", new Concatenation( new Singleton(">>"), new Concatenation( new Repetition( new Union( // either not a closing bracket.. new NonMatch("<"), // or one closing bracket but not two of them new Concatenation( new Singleton("<"), new NonMatch("<")) ), 0, INFINITY), new Singleton("<<") ))); /** * 19.3 Terminals from section 3.10.2: Floating-Point Literal */ put("FLOATING_POINT_LITERAL", new Union( /** * [[:digit:]]+\.[[:digit:]]*([eE][-+]?[[:digit:]]+)?[fFdD]? */ new Concatenation( new Repetition(PosixClass.digit(), 1, INFINITY), new Concatenation( new Singleton("."), new Concatenation( new Repetition(PosixClass.digit(), allowShortFloatingPoints ? 0 : 1, INFINITY), new Concatenation( new Repetition(new Concatenation( new Match("eE"), new Concatenation( new Repetition(new Match("-+"), 0, 1), new Repetition(PosixClass.digit(), 1, INFINITY) )), 0, 1), new Repetition(new Match("fFdD"), 0, 1) )))), new Union( /** * \.[[:digit:]]+([eE][-+]?[[:digit:]]+)?[fFdD]? */ new Concatenation( new Singleton("."), new Concatenation( new Repetition(PosixClass.digit(), 1, INFINITY), new Concatenation( new Repetition(new Concatenation( new Match("eE"), new Concatenation( new Repetition(new Match("-+"), 0, 1), new Repetition(PosixClass.digit(), 1, INFINITY) )), 0, 1), new Repetition(new Match("fFdD"), 0, 1) ))), new Union( /** * [[:digit:]]+[eE][-+]?[[:digit:]]+[fFdD]? */ new Concatenation( new Repetition(PosixClass.digit(), 1, INFINITY), new Concatenation( new Match("eE"), new Concatenation( new Repetition(new Match("-+"), 0, 1), new Concatenation( new Repetition(PosixClass.digit(), 1, INFINITY), new Repetition(new Match("fFdD"), 0, 1) )))), /** * [[:digit:]]+([eE][-+]?[[:digit:]]+)?[fFdD] */ new Concatenation( new Repetition(PosixClass.digit(), 1, INFINITY), new Concatenation( new Repetition(new Concatenation( new Match("eE"), new Concatenation( new Repetition(new Match("-+"), 0, 1), new Repetition(PosixClass.digit(), 1, INFINITY) )), 0, 1), new Match("fFdD") )) )))); } } // class MyLexicon }