processinginstruction.java

一个用于搜索本地文件内容的小型搜索引擎

    /**
     * This will set a pseudo attribute with the given name and value.
     * If the PI data is not already in a pseudo-attribute format, this will
     * replace the existing data.
     *
     * @param name <code>String</code> name of pair.
     * @param value <code>String</code> value for pair.
     * @return <code>ProcessingInstruction</code> this PI modified.
     */
    public ProcessingInstruction setPseudoAttribute(String name, String value) {
        String reason = Verifier.checkProcessingInstructionData(name);
        if (reason != null) {
            throw new IllegalDataException(name, reason);
        }

        reason = Verifier.checkProcessingInstructionData(value);
        if (reason != null) {
            throw new IllegalDataException(value, reason);
        }

        this.mapData.put(name, value);
        this.rawData = toString(mapData);
        return this;
    }


    /**
     * This will remove the pseudo attribute with the specified name.
     *
     * @param name name of pseudo attribute to remove
     * @return <code>boolean</code> - whether the requested
     *         instruction was removed.
     */
    public boolean removePseudoAttribute(String name) {
        if ((mapData.remove(name)) != null) {
            rawData = toString(mapData);
            return true;
        }

        return false;
    }

    /**
     * This will convert the Map to a string representation.
     *
     * @param mapData <code>Map</code> PI data to convert
     * @return a string representation of the Map as appropriate for a PI
     */
    private String toString(Map mapData) {
        StringBuffer rawData = new StringBuffer();

        Iterator i = mapData.keySet().iterator();
        while (i.hasNext()) {
            String name = (String)i.next();
            String value = (String)mapData.get(name);
            rawData.append(name)
                   .append("=\"")
                   .append(value)
                   .append("\" ");
        }
        // Remove last space, if we did any appending
        if (rawData.length() > 0) {
            rawData.setLength(rawData.length() - 1);
        }

        return rawData.toString();
    }

    /**
     * This will parse and load the instructions for the PI.
     * This is separated to allow it to occur once and then be reused.
     */
    private Map parseData(String rawData) {
        // The parsing here is done largely "by hand" which means the code
        // gets a little tricky/messy.  The following conditions should
        // now be handled correctly:
        //   <?pi href="http://hi/a=b"?>        Reads OK
        //   <?pi href = 'http://hi/a=b' ?>     Reads OK
        //   <?pi href\t = \t'http://hi/a=b'?>  Reads OK
        //   <?pi href  =  "http://hi/a=b"?>    Reads OK
        //   <?pi?>                             Empty Map
        //   <?pi id=22?>                       Empty Map
        //   <?pi id='22?>                      Empty Map

        Map data = new HashMap();

        // System.out.println("rawData: " + rawData);

        // The inputData variable holds the part of rawData left to parse
        String inputData = rawData.trim();

        // Iterate through the remaining inputData string
        while (!inputData.trim().equals("")) {
            //System.out.println("parseData() looking at: " + inputData);

            // Search for "name =", "name=" or "name1 name2..."
            String name = "";
            String value = "";
            int startName = 0;
            char previousChar = inputData.charAt(startName);
            int pos = 1;
            for (; pos<inputData.length(); pos++) {
                char currentChar = inputData.charAt(pos);
                if (currentChar == '=') {
                    name = inputData.substring(startName, pos).trim();
                    // Get the boundaries on the quoted string
                    // We use boundaries so we know where to start next
                    int[] bounds = extractQuotedString(
                                     inputData.substring(pos+1));
                    // A null value means a parse error and we return empty!
                    if (bounds == null) {
                        return new HashMap();
                    }
                    value = inputData.substring(bounds[0]+pos+1,
                                                bounds[1]+pos+1);
                    pos += bounds[1] + 1;  // skip past value
                    break;
                }
                else if (Character.isWhitespace(previousChar)
                          && !Character.isWhitespace(currentChar)) {
                    startName = pos;
                }

                previousChar = currentChar;
            }

            // Remove the first pos characters; they have been processed
            inputData = inputData.substring(pos);

            // System.out.println("Extracted (name, value) pair: ("
            //                          + name + ", '" + value+"')");

            // If both a name and a value have been found, then add
            // them to the data Map
            if (name.length() > 0 && value != null) {
                //if (data.containsKey(name)) {
                    // A repeat, that's a parse error, so return a null map
                    //return new HashMap();
                //}
                //else {
                    data.put(name, value);
                //}
            }
        }

        return data;
    }

    /**
     * This is a helper routine, only used by parseData, to extract a
     * quoted String from the input parameter, rawData. A quoted string
     * can use either single or double quotes, but they must match up.
     * A singly quoted string can contain an unbalanced amount of double
     * quotes, or vice versa. For example, the String "JDOM's the best"
     * is legal as is 'JDOM"s the best'.
     *
     * @param rawData the input string from which a quoted string is to
     *                be extracted.
     * @return the first quoted string encountered in the input data. If
     *         no quoted string is found, then the empty string, "", is
     *         returned.
     * @see #parseData
     */
    private static int[] extractQuotedString(String rawData) {
        // Remembers whether we're actually in a quoted string yet
        boolean inQuotes = false;

        // Remembers which type of quoted string we're in
        char quoteChar = '"';

        // Stores the position of the first character inside
        //  the quoted string (i.e. the start of the return string)
        int start = 0;

        // Iterate through the input string looking for the start
        // and end of the quoted string
        for (int pos=0; pos < rawData.length(); pos++) {
            char currentChar = rawData.charAt(pos);
            if (currentChar=='"' || currentChar=='\'') {
                if (!inQuotes) {
                    // We're entering a quoted string
                    quoteChar = currentChar;
                    inQuotes = true;
                    start = pos+1;
                }
                else if (quoteChar == currentChar) {
                    // We're leaving a quoted string
                    inQuotes = false;
                    return new int[] { start, pos };
                }
                // Otherwise we've encountered a quote
                // inside a quote, so just continue
            }
        }

        return null;
    }

    /**
     * This returns a <code>String</code> representation of the
     * <code>ProcessingInstruction</code>, suitable for debugging. If the XML
     * representation of the <code>ProcessingInstruction</code> is desired,
     * {@link org.jdom.output.XMLOutputter#outputString(ProcessingInstruction)}
     * should be used.
     *
     * @return <code>String</code> - information about the
     *         <code>ProcessingInstruction</code>
     */
    public String toString() {
        return new StringBuffer()
            .append("[ProcessingInstruction: ")
            .append(new org.jdom.output.XMLOutputter().outputString(this))
            .append("]")
            .toString();
    }

    /**
     * This will return a clone of this <code>ProcessingInstruction</code>.
     *
     * @return <code>Object</code> - clone of this
     * <code>ProcessingInstruction</code>.
     */
    public Object clone() {
        ProcessingInstruction pi = (ProcessingInstruction) super.clone();

        // target and rawdata are immutable and references copied by
        // Object.clone()

        // Create a new Map object for the clone (since Map isn't Cloneable)
        if (mapData != null) {
            pi.mapData = parseData(rawData);
        }
        return pi;
    }
}