xpath.java

java1.6众多例子参考

        //
        // Data
        //

        /** Symbol table. */
        private SymbolTable fSymbolTable;

        // symbols

        private static final String fAndSymbol = "and".intern();
        private static final String fOrSymbol = "or".intern();
        private static final String fModSymbol = "mod".intern();
        private static final String fDivSymbol = "div".intern();

        private static final String fCommentSymbol = "comment".intern();
        private static final String fTextSymbol = "text".intern();
        private static final String fPISymbol = "processing-instruction".intern();
        private static final String fNodeSymbol = "node".intern();

        private static final String fAncestorSymbol = "ancestor".intern();
        private static final String fAncestorOrSelfSymbol = "ancestor-or-self".intern();
        private static final String fAttributeSymbol = "attribute".intern();
        private static final String fChildSymbol = "child".intern();
        private static final String fDescendantSymbol = "descendant".intern();
        private static final String fDescendantOrSelfSymbol = "descendant-or-self".intern();
        private static final String fFollowingSymbol = "following".intern();
        private static final String fFollowingSiblingSymbol = "following-sibling".intern();
        private static final String fNamespaceSymbol = "namespace".intern();
        private static final String fParentSymbol = "parent".intern();
        private static final String fPrecedingSymbol = "preceding".intern();
        private static final String fPrecedingSiblingSymbol = "preceding-sibling".intern();
        private static final String fSelfSymbol = "self".intern();

        //
        // Constructors
        //

        /** Constructs an XPath expression scanner. */
        public Scanner(SymbolTable symbolTable) {

            // save pool and tokens
            fSymbolTable = symbolTable;

        } // <init>(SymbolTable)

        /**
         *
         */
        public boolean scanExpr(SymbolTable symbolTable,
                                XPath.Tokens tokens, String data,
                                int currentOffset, int endOffset)
            throws XPathException {

            int nameOffset;
            String nameHandle, prefixHandle;
            boolean starIsMultiplyOperator = false;
            int ch;

            while (true) {
                if (currentOffset == endOffset) {
                    break;
                }
                ch = data.charAt(currentOffset);
                //
                // [39] ExprWhitespace ::= S
                //
                while (ch == ' ' || ch == 0x0A || ch == 0x09 || ch == 0x0D) {
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    ch = data.charAt(currentOffset);
                }
                if (currentOffset == endOffset) {
                    break;
                }
                //
                // [28] ExprToken ::= '(' | ')' | '[' | ']' | '.' | '..' | '@' | ',' | '::'
                //                  | NameTest | NodeType | Operator | FunctionName
                //                  | AxisName | Literal | Number | VariableReference
                //
                byte chartype = (ch >= 0x80) ? CHARTYPE_NONASCII : fASCIICharMap[ch];
                switch (chartype) {
                case CHARTYPE_OPEN_PAREN:       // '('
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_OPEN_PAREN);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_CLOSE_PAREN:      // ')'
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_CLOSE_PAREN);
                    starIsMultiplyOperator = true;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_OPEN_BRACKET:     // '['
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_OPEN_BRACKET);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_CLOSE_BRACKET:    // ']'
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_CLOSE_BRACKET);
                    starIsMultiplyOperator = true;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                //
                // [30] Number ::= Digits ('.' Digits?)? | '.' Digits
                //                                         ^^^^^^^^^^
                //
                case CHARTYPE_PERIOD:           // '.', '..' or '.' Digits
                    if (currentOffset + 1 == endOffset) {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_PERIOD);
                        starIsMultiplyOperator = true;
                        currentOffset++;
                        break;
                    }
                    ch = data.charAt(currentOffset + 1);
                    if (ch == '.') {            // '..'
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_DOUBLE_PERIOD);
                        starIsMultiplyOperator = true;
                        currentOffset += 2;
                    } else if (ch >= '0' && ch <= '9') {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_NUMBER);
                        starIsMultiplyOperator = true;
                        currentOffset = scanNumber(tokens, data, endOffset, currentOffset/*, encoding*/);
                    } else if (ch == '/') {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_PERIOD);
                        starIsMultiplyOperator = true;
                        currentOffset++;
                    } else if (ch == '|') {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_PERIOD);
                        starIsMultiplyOperator = true;
                        currentOffset++;
                        break;
                    } else if (ch == ' ' || ch == 0x0A || ch == 0x09 || ch == 0x0D) {
                        // this is legal if the next token is non-existent or |
                        do {
                            if (++currentOffset == endOffset) {
                                break;
                            }
                            ch = data.charAt(currentOffset);
                        } while (ch == ' ' || ch == 0x0A || ch == 0x09 || ch == 0x0D); 
                        if (currentOffset == endOffset || ch == '|') {
                            addToken(tokens, XPath.Tokens.EXPRTOKEN_PERIOD);
                            starIsMultiplyOperator = true;
                            break;
                        }
                        throw new XPathException ("c-general-xpath");
                    } else {                    // '.'
                        throw new XPathException ("c-general-xpath");
                    }
                    if (currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_ATSIGN:           // '@'
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_ATSIGN);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_COMMA:            // ','
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_COMMA);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_COLON:            // '::'
                    if (++currentOffset == endOffset) {
                // System.out.println("abort 1a");
                        return false; // REVISIT
                    }
                    ch = data.charAt(currentOffset);
                    if (ch != ':') {
                // System.out.println("abort 1b");
                        return false; // REVISIT
                    }
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_DOUBLE_COLON);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_SLASH:            // '/' and '//'
                    if (++currentOffset == endOffset) {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_SLASH);
                        starIsMultiplyOperator = false;
                        break;
                    }
                    ch = data.charAt(currentOffset);
                    if (ch == '/') { // '//'
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_DOUBLE_SLASH);
                        starIsMultiplyOperator = false;
                        if (++currentOffset == endOffset) {
                            break;
                        }
                    } else {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_SLASH);
                        starIsMultiplyOperator = false;
                    }
                    break;
                case CHARTYPE_UNION:            // '|'
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_UNION);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_PLUS:             // '+'
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_PLUS);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_MINUS:            // '-'
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_MINUS);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_EQUAL:            // '='
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_EQUAL);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_EXCLAMATION:      // '!='
                    if (++currentOffset == endOffset) {
                // System.out.println("abort 2a");
                        return false; // REVISIT
                    }
                    ch = data.charAt(currentOffset);
                    if (ch != '=') {
                // System.out.println("abort 2b");
                        return false; // REVISIT
                    }
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_NOT_EQUAL);
                    starIsMultiplyOperator = false;
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                case CHARTYPE_LESS: // '<' and '<='
                    if (++currentOffset == endOffset) {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_LESS);
                        starIsMultiplyOperator = false;
                        break;
                    }
                    ch = data.charAt(currentOffset);
                    if (ch == '=') { // '<='
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_LESS_EQUAL);
                        starIsMultiplyOperator = false;
                        if (++currentOffset == endOffset) {
                            break;
                        }
                    } else {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_LESS);
                        starIsMultiplyOperator = false;
                    }
                    break;
                case CHARTYPE_GREATER: // '>' and '>='
                    if (++currentOffset == endOffset) {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_GREATER);
                        starIsMultiplyOperator = false;
                        break;
                    }
                    ch = data.charAt(currentOffset);
                    if (ch == '=') { // '>='
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_GREATER_EQUAL);
                        starIsMultiplyOperator = false;
                        if (++currentOffset == endOffset) {
                            break;
                        }
                    } else {
                        addToken(tokens, XPath.Tokens.EXPRTOKEN_OPERATOR_GREATER);
                        starIsMultiplyOperator = false;
                    }
                    break;
                //
                // [29] Literal ::= '"' [^"]* '"' | "'" [^']* "'"
                //
                case CHARTYPE_QUOTE:            // '\"' or '\''
                    int qchar = ch;
                    if (++currentOffset == endOffset) {
                // System.out.println("abort 2c");
                        return false; // REVISIT
                    }
                    ch = data.charAt(currentOffset);
                    int litOffset = currentOffset;
                    while (ch != qchar) {
                        if (++currentOffset == endOffset) {
                // System.out.println("abort 2d");
                            return false; // REVISIT
                        }
                        ch = data.charAt(currentOffset);
                    }
                    int litLength = currentOffset - litOffset;
                    addToken(tokens, XPath.Tokens.EXPRTOKEN_LITERAL);
                    starIsMultiplyOperator = true;
                    tokens.addToken(symbolTable.addSymbol(data.substring(litOffset, litOffset + litLength)));
                    if (++currentOffset == endOffset) {
                        break;
                    }
                    break;
                //
                // [30] Number ::= Digits ('.' Digits?)? | '.' Digits
                // [31] Digits ::= [0-9]+