/**
* Static service to get a term from its string representation, providing a specific operator
* manager
*/
public static Term parseSingleTerm(String st, OperatorManager op) throws InvalidTermException {
try {
Parser p = new Parser(op, st);
Token t = p.tokenizer.readToken();
if (t.isEOF()) throw new InvalidTermException("Term starts with EOF");
p.tokenizer.unreadToken(t);
Term term = p.expr(false);
if (term == null) throw new InvalidTermException("Term is null");
if (!p.tokenizer.readToken().isEOF())
throw new InvalidTermException("The entire string could not be read as one term");
term.resolveTerm();
return term;
} catch (IOException ex) {
throw new InvalidTermException("An I/O error occured");
}
}
/**
* Parses and returns a valid 'leftside' of an expression. If the left side starts with a prefix,
* it consumes other expressions with a lower priority than itself. If the left side does not have
* a prefix it must be an expr0.
*
* @param commaIsEndMarker used when the leftside is part of and argument list of expressions
* @param maxPriority operators with a higher priority than this will effectivly end the
* expression
* @return a wrapper of: 1. term correctly structured and 2. the priority of its root operator
* @throws InvalidTermException
*/
private IdentifiedTerm parseLeftSide(boolean commaIsEndMarker, int maxPriority)
throws InvalidTermException, IOException {
// 1. prefix expression
Token f = tokenizer.readToken();
if (f.isOperator(commaIsEndMarker)) {
int FX = opManager.opPrio(f.seq, "fx");
int FY = opManager.opPrio(f.seq, "fy");
if (f.seq.equals("-")) {
Token t = tokenizer.readToken();
if (t.isNumber())
/*Michele Castagna 06/2011*/
// return new IdentifiedTerm(0, Parser.createNumber("-" + t.seq));
return identifyTerm(0, Parser.createNumber("-" + t.seq), tokenStart);
/**/
else tokenizer.unreadToken(t);
}
// check that no operator has a priority higher than permitted
if (FY > maxPriority) FY = -1;
if (FX > maxPriority) FX = -1;
// FX has priority over FY
boolean haveAttemptedFX = false;
if (FX >= FY && FX >= OperatorManager.OP_LOW) {
IdentifiedTerm found = exprA(FX - 1, commaIsEndMarker); // op(fx, n) exprA(n - 1)
if (found != null)
/*Castagna 06/2011*/
// return new IdentifiedTerm(FX, new Struct(f.seq, found.result));
return identifyTerm(FX, new Struct(f.seq, found.result), tokenStart);
/**/
else haveAttemptedFX = true;
}
// FY has priority over FX, or FX has failed
if (FY >= OperatorManager.OP_LOW) {
IdentifiedTerm found =
exprA(FY, commaIsEndMarker); // op(fy,n) exprA(1200) or op(fy,n) exprA(n)
if (found != null)
/*Castagna 06/2011*/
// return new IdentifiedTerm(FY, new Struct(f.seq, found.result));
return identifyTerm(FY, new Struct(f.seq, found.result), tokenStart);
/**/
}
// FY has priority over FX, but FY failed
if (!haveAttemptedFX && FX >= OperatorManager.OP_LOW) {
IdentifiedTerm found = exprA(FX - 1, commaIsEndMarker); // op(fx, n) exprA(n - 1)
if (found != null)
/*Castagna 06/2011*/
// return new IdentifiedTerm(FX, new Struct(f.seq, found.result));
return identifyTerm(FX, new Struct(f.seq, found.result), tokenStart);
/**/
}
}
tokenizer.unreadToken(f);
// 2. expr0
return new IdentifiedTerm(0, expr0());
}
Token readNextToken() throws IOException, InvalidTermException {
int typea;
String svala;
if (pushBack2 != null) {
typea = pushBack2.typea;
svala = pushBack2.svala;
pushBack2 = null;
} else {
typea = super.nextToken();
svala = sval;
}
// skips whitespace
// could be simplified if lookahead for blank space in functors wasn't necessary
// and if '.' in numbers could be written with blank space
while (Tokenizer.isWhite(typea)) {
typea = super.nextToken();
svala = sval;
}
// skips single line comments
// could be simplified if % was not a legal character in quotes
if (typea == '%') {
do {
typea = super.nextToken();
} while (typea != '\r' && typea != '\n' && typea != TT_EOF);
pushBack(); // pushes back \r or \n. These are whitespace, so when readNextToken() finds them,
// they are marked as whitespace
return readNextToken();
}
// skips /* comments */
if (typea == '/') {
int typeb = super.nextToken();
if (typeb == '*') {
do {
typea = typeb;
typeb = super.nextToken();
} while (typea != '*' || typeb != '/');
return readNextToken();
} else {
pushBack();
}
}
// syntactic charachters
if (typea == TT_EOF) return new Token("", Tokenizer.EOF);
if (typea == '(') return new Token("(", Tokenizer.LPAR);
if (typea == ')') return new Token(")", Tokenizer.RPAR);
if (typea == '{') return new Token("{", Tokenizer.LBRA2);
if (typea == '}') return new Token("}", Tokenizer.RBRA2);
if (typea == '[') return new Token("[", Tokenizer.LBRA);
if (typea == ']') return new Token("]", Tokenizer.RBRA);
if (typea == '|') return new Token("|", Tokenizer.BAR);
if (typea == '!') return new Token("!", Tokenizer.ATOM);
if (typea == ',') return new Token(",", Tokenizer.OPERATOR);
if (typea
== '.') { // check that '.' as end token is followed by a layout character, see ISO Standard
// 6.4.8 endnote
int typeb = super.nextToken();
if (Tokenizer.isWhite(typeb) || typeb == '%' || typeb == StreamTokenizer.TT_EOF)
return new Token(".", Tokenizer.END);
else pushBack();
}
boolean isNumber = false;
// variable, atom or number
if (typea == TT_WORD) {
char firstChar = svala.charAt(0);
// variable
if (Character.isUpperCase(firstChar) || '_' == firstChar)
return new Token(svala, Tokenizer.VARIABLE);
else if (firstChar >= '0'
&& firstChar <= '9') // all words starting with 0 or 9 must be a number
isNumber = true; // set type to number and handle later
else { // otherwise, it must be an atom (or wrong)
int typeb = super.nextToken(); // lookahead 1 to identify what type of atom
pushBack(); // this does not skip whitespaces, only readNext does so.
if (typeb == '(') return new Token(svala, Tokenizer.ATOM | Tokenizer.FUNCTOR);
if (Tokenizer.isWhite(typeb)) return new Token(svala, Tokenizer.ATOM | Tokenizer.OPERATOR);
return new Token(svala, Tokenizer.ATOM);
}
}
// quotes
if (typea == '\'' || typea == '\"' || typea == '`') {
int qType = typea;
StringBuffer quote = new StringBuffer();
while (true) { // run through entire quote and added body to quote buffer
typea = super.nextToken();
svala = sval;
// continuation escape sequence
if (typea == '\\') {
int typeb = super.nextToken();
if (typeb == '\n') // continuation escape sequence marker \\n
continue;
if (typeb == '\r') {
int typec = super.nextToken();
if (typec == '\n') continue; // continuation escape sequence marker \\r\n
pushBack();
continue; // continuation escape sequence marker \\r
}
pushBack(); // pushback typeb
}
// double '' or "" or ``
if (typea == qType) {
int typeb = super.nextToken();
if (typeb == qType) { // escaped '' or "" or ``
quote.append((char) qType);
continue;
} else {
pushBack();
break; // otherwise, break on single quote
}
}
if (typea == '\n' || typea == '\r')
throw new InvalidTermException(
"line break in quote not allowed (unless they are escaped \\ first)");
if (svala != null) quote.append(svala);
else quote.append((char) typea);
}
String quoteBody = quote.toString();
qType = qType == '\'' ? SQ_SEQUENCE : qType == '\"' ? DQ_SEQUENCE : SQ_SEQUENCE;
if (qType == SQ_SEQUENCE) {
if (Parser.isAtom(quoteBody)) qType = ATOM;
int typeb = super.nextToken(); // lookahead 1 to identify what type of quote
pushBack(); // nextToken() does not skip whitespaces, only readNext does so.
if (typeb == '(') return new Token(quoteBody, qType | FUNCTOR);
}
return new Token(quoteBody, qType);
}
// symbols
if (Arrays.binarySearch(Tokenizer.GRAPHIC_CHARS, (char) typea) >= 0) {
// the symbols are parsed individually by the super.nextToken(), so accumulate symbollist
StringBuffer symbols = new StringBuffer();
int typeb = typea;
// String svalb = null;
while (Arrays.binarySearch(Tokenizer.GRAPHIC_CHARS, (char) typeb) >= 0) {
symbols.append((char) typeb);
typeb = super.nextToken();
// svalb = sval;
}
pushBack();
// special symbols: unary + and unary -
// try {
// if (symbols.length() == 1 && typeb == TT_WORD &&
// java.lang.Long.parseLong(svalb) > 0) {
// if (typea == '+') //todo, issue of handling +
// and -. I don't think this is ISO..
// return readNextToken(); //skips + and returns the next
// number
// if (typea == '-') {
// Token t = readNextToken(); //read the next number
// t.seq = "-" + t.seq; //add minus to value
// return t; //return token
// }
// } //ps. the reason why the
// number isn't returned right away, but through nextToken(), is because the number might be
// for instance a float
// } catch (NumberFormatException e) {
// }
return new Token(symbols.toString(), Tokenizer.OPERATOR);
}
// numbers: 1. integer, 2. float
if (isNumber) {
try { // the various parseInt checks will throw exceptions when parts of numbers are written
// illegally
// 1.a. complex integers
if (svala.startsWith("0")) {
if (svala.indexOf('b') == 1)
return new Token(
"" + java.lang.Long.parseLong(svala.substring(2), 2),
Tokenizer.INTEGER); // try binary
if (svala.indexOf('o') == 1)
return new Token(
"" + java.lang.Long.parseLong(svala.substring(2), 8),
Tokenizer.INTEGER); // try octal
if (svala.indexOf('x') == 1)
return new Token(
"" + java.lang.Long.parseLong(svala.substring(2), 16),
Tokenizer.INTEGER); // try hex
}
// lookahead 1
int typeb = super.nextToken();
String svalb = sval;
// 1.b ordinary integers
if (typeb != '.' && typeb != '\'') { // i.e. not float or character constant
pushBack(); // lookahead 0
return new Token("" + java.lang.Long.parseLong(svala), Tokenizer.INTEGER);
}
// 1.c character code constant
if (typeb == '\'' && "0".equals(svala)) {
int typec = super.nextToken(); // lookahead 2
String svalc = sval;
int intVal;
if ((intVal = isCharacterCodeConstantToken(typec, svalc)) != -1)
return new Token("" + intVal, Tokenizer.INTEGER);
// this is an invalid character code constant int
throw new InvalidTermException(
"Character code constant starting with 0'<X> at line: "
+ super.lineno()
+ " cannot be recognized.");
}
// 2.a check that the value of the word prior to period is a valid long
java.lang.Long.parseLong(svala); // throws an exception if not
// 2.b first int is followed by a period
if (typeb != '.')
throw new InvalidTermException(
"A number starting with 0-9 cannot be rcognized as an int and does not have a fraction '.' at line: "
+ super.lineno());
// lookahead 2
int typec = super.nextToken();
String svalc = sval;
// 2.c check that the next token after '.' is a possible fraction
if (typec != TT_WORD) { // if its not, the period is an End period
pushBack(); // pushback 1 the token after period
pushBack2 = new PushBack(typeb, svalb); // pushback 2 the period token
return new Token(svala, INTEGER); // return what must be an int
}
// 2.d checking for exponent
int exponent = svalc.indexOf("E");
if (exponent == -1) exponent = svalc.indexOf("e");
if (exponent >= 1) { // the float must have a valid exponent
if (exponent == svalc.length() - 1) { // the exponent must be signed exponent
int typeb2 = super.nextToken();
if (typeb2 == '+' || typeb2 == '-') {
int typec2 = super.nextToken();
String svalc2 = sval;
if (typec2 == TT_WORD) {
// verify the remaining parts of the float and return
java.lang.Long.parseLong(svalc.substring(0, exponent));
java.lang.Integer.parseInt(svalc2);
return new Token(svala + "." + svalc + (char) typeb2 + svalc2, Tokenizer.FLOAT);
}
}
}
}
// 2.e verify lastly that ordinary floats and unsigned exponent floats are Java legal and
// return them
java.lang.Double.parseDouble(svala + "." + svalc);
return new Token(svala + "." + svalc, Tokenizer.FLOAT);
} catch (NumberFormatException e) {
// TODO return more info on what was wrong with the number given
throw new InvalidTermException(
"A term starting with 0-9 cannot be parsed as a number at line: " + lineno());
}
}
throw new InvalidTermException("Unknown Unicode character: " + typea + " (" + svala + ")");
}
/**
* exprA(0) ::= integer | float | variable | atom | atom( exprA(1200) { , exprA(1200) }* ) | '['
* exprA(1200) { , exprA(1200) }* [ | exprA(1200) ] ']' | '{' [ exprA(1200) ] '}' | '('
* exprA(1200) ')'
*/
private Term expr0() throws InvalidTermException, IOException {
Token t1 = tokenizer.readToken();
/*Castagna 06/2011*/
/*
if (t1.isType(Tokenizer.INTEGER))
return Parser.parseInteger(t1.seq); //todo moved method to Number
if (t1.isType(Tokenizer.FLOAT))
return Parser.parseFloat(t1.seq); //todo moved method to Number
if (t1.isType(Tokenizer.VARIABLE))
return new Var(t1.seq); //todo switched to use the internal check for "_" in Var(String)
*/
int tempStart = tokenizer.tokenStart();
if (t1.isType(Tokenizer.INTEGER)) {
Term i = Parser.parseInteger(t1.seq);
map(i, tokenizer.tokenStart());
return i; // todo moved method to Number
}
if (t1.isType(Tokenizer.FLOAT)) {
Term f = Parser.parseFloat(t1.seq);
map(f, tokenizer.tokenStart());
return f; // todo moved method to Number
}
if (t1.isType(Tokenizer.VARIABLE)) {
Term v = new Var(t1.seq);
map(v, tokenizer.tokenStart());
return v; // todo switched to use the internal check for "_" in Var(String)
}
/**/
if (t1.isType(Tokenizer.ATOM)
|| t1.isType(Tokenizer.SQ_SEQUENCE)
|| t1.isType(Tokenizer.DQ_SEQUENCE)) {
if (!t1.isFunctor())
/*Castagna 06/2011*/
{
// return new Struct(t1.seq);
Term f = new Struct(t1.seq);
map(f, tokenizer.tokenStart());
return f;
}
/**/
String functor = t1.seq;
Token t2 = tokenizer.readToken(); // reading left par
if (!t2.isType(Tokenizer.LPAR))
throw new InvalidTermException(
"Something identified as functor misses its first left parenthesis"); // todo check can
// be skipped
LinkedList<Term> a = expr0_arglist(); // reading arguments
Token t3 = tokenizer.readToken();
if (t3.isType(Tokenizer.RPAR)) // reading right par
/*Castagna 06/2011*/
{
// return new Struct(functor, a);
Term c = new Struct(functor, a);
map(c, tempStart);
return c;
}
/**/
/*Castagna 06/2011*/
// throw new InvalidTermException("Missing right parenthesis: ("+a + " -> here <-");
throw new InvalidTermException(
"Missing right parenthesis '(" + a + "' -> here <-",
tokenizer.offsetToRowColumn(getCurrentOffset())[0],
tokenizer.offsetToRowColumn(getCurrentOffset())[1] - 1);
/**/
}
if (t1.isType(Tokenizer.LPAR)) {
Term term = expr(false);
if (tokenizer.readToken().isType(Tokenizer.RPAR)) return term;
/*Castagna 06/2011*/
// throw new InvalidTermException("Missing right parenthesis: ("+term + " -> here <-");
throw new InvalidTermException(
"Missing right parenthesis '(" + term + "' -> here <-",
tokenizer.offsetToRowColumn(getCurrentOffset())[0],
tokenizer.offsetToRowColumn(getCurrentOffset())[1] - 1);
/**/
}
if (t1.isType(Tokenizer.LBRA)) {
Token t2 = tokenizer.readToken();
if (t2.isType(Tokenizer.RBRA)) return new Struct();
tokenizer.unreadToken(t2);
Term term = expr0_list();
if (tokenizer.readToken().isType(Tokenizer.RBRA)) return term;
/*Castagna 06/2011*/
// throw new InvalidTermException("Missing right bracket: ["+term + " -> here <-");
throw new InvalidTermException(
"Missing right bracket '[" + term + " ->' here <-",
tokenizer.offsetToRowColumn(getCurrentOffset())[0],
tokenizer.offsetToRowColumn(getCurrentOffset())[1] - 1);
/**/
}
if (t1.isType(Tokenizer.LBRA2)) {
Token t2 = tokenizer.readToken();
if (t2.isType(Tokenizer.RBRA2))
/*Castagna 06/2011*/
{
// return new Struct("{}");
Term b = new Struct("{}");
map(b, tempStart);
return b;
}
/**/
tokenizer.unreadToken(t2);
Term arg = expr(false);
t2 = tokenizer.readToken();
if (t2.isType(Tokenizer.RBRA2))
/*Castagna 06/2011*/
{
// return new Struct("{}", arg);
Term b = new Struct("{}", arg);
map(b, tempStart);
return b;
}
/*Castagna 06/2011*/
// throw new InvalidTermException("Missing right braces: {"+arg + " -> here <-");
throw new InvalidTermException(
"Missing right braces '{" + arg + "' -> here <-",
tokenizer.offsetToRowColumn(getCurrentOffset())[0],
tokenizer.offsetToRowColumn(getCurrentOffset())[1] - 1);
/**/
}
/*Castagna 06/2011*/
// throw new InvalidTermException("The following token could not be identified: "+t1.seq);
throw new InvalidTermException(
"Unexpected token '" + t1.seq + "'",
tokenizer.offsetToRowColumn(getCurrentOffset())[0],
tokenizer.offsetToRowColumn(getCurrentOffset())[1] - 1);
/**/
}
