crawlsettingssaxsource.java

最强的爬虫工程

        while (it.hasNext()) {
            Object limit = it.next();
            if (limit instanceof TimespanCriteria) {
                AttributesImpl timeSpan = new AttributesImpl();
                timeSpan.addAttribute(nsu,
                        XMLSettingsHandler.XML_ATTRIBUTE_FROM,
                        XMLSettingsHandler.XML_ATTRIBUTE_FROM, nsu,
                        ((TimespanCriteria) limit).getFrom());
                timeSpan.addAttribute(nsu, XMLSettingsHandler.XML_ATTRIBUTE_TO,
                        XMLSettingsHandler.XML_ATTRIBUTE_TO, nsu,
                        ((TimespanCriteria) limit).getTo());
                writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_TIMESPAN, "",
                        timeSpan, indent + 2 * indentAmount);
            } else if (limit instanceof PortnumberCriteria) {
                writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_PORTNUMBER,
                        ((PortnumberCriteria) limit).getPortNumber(), nullAtts,
                        indent + 2 * indentAmount);
            } else if (limit instanceof RegularExpressionCriteria) {
                writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_URIMATCHES,
                        ((RegularExpressionCriteria) limit).getRegexp(), nullAtts,
                        indent + 2 * indentAmount);
            }
        }

        handler.ignorableWhitespace(indentArray, 0, indent);
        handler.endElement(nsu, XMLSettingsHandler.XML_ELEMENT_LIMITS,
                XMLSettingsHandler.XML_ELEMENT_LIMITS);

    }

    private void parseMetaData(int indent) throws SAXException {
        // Write meta information
        Attributes nullAtts = new AttributesImpl();
        handler.ignorableWhitespace(indentArray, 0, indent);
        handler.startElement(nsu, XMLSettingsHandler.XML_ELEMENT_META,
                XMLSettingsHandler.XML_ELEMENT_META, nullAtts);

        // Write settings name
        writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_NAME, settings
                .getName(), null, indent + indentAmount);

        // Write settings description
        writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_DESCRIPTION, settings
                .getDescription(), null, indent + indentAmount);

        // Write settings operator
        writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_OPERATOR, settings
                .getOperator(), null, indent + indentAmount);

        // Write settings description
        writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_ORGANIZATION, settings
                .getOrganization(), null, indent + indentAmount);

        // Write settings description
        writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_AUDIENCE, settings
                .getAudience(), null, indent + indentAmount);

        // Write file date
        String dateStamp = ArchiveUtils.get14DigitDate();
        writeSimpleElement(XMLSettingsHandler.XML_ELEMENT_DATE, dateStamp,
                null, indent + indentAmount);
        try {
            settings.setLastSavedTime(ArchiveUtils.parse14DigitDate(dateStamp));
        } catch (ParseException e) {
            // Should never happen since we just created it. If this exception
            // is thrown, then there is a bug in ArchiveUtils.
            e.printStackTrace();
        }

        handler.ignorableWhitespace(indentArray, 0, indent);
        handler.endElement(nsu, XMLSettingsHandler.XML_ELEMENT_META,
                XMLSettingsHandler.XML_ELEMENT_META);
    }

    /**
     * Create SAX events from a {@link ComplexType}.
     *
     * @param complexType the object to creat SAX events from.
     * @param indent the indentation amount for prettyprinting XML.
     * @throws SAXException is thrown if an error occurs.
     */
    private void parseComplexType(ComplexType complexType, int indent)
            throws SAXException {
        if (complexType.isTransient()) {
            return;
        }
        MBeanInfo mbeanInfo = complexType.getMBeanInfo(settings);
        String objectElement = resolveElementName(complexType);
        AttributesImpl atts = new AttributesImpl();
        atts.addAttribute(nsu, XMLSettingsHandler.XML_ATTRIBUTE_NAME,
                XMLSettingsHandler.XML_ATTRIBUTE_NAME, nsu, complexType
                        .getName());
        if (objectElement == XMLSettingsHandler.XML_ELEMENT_NEW_OBJECT) {
            // Only 'newObject' elements have a class attribute
            atts.addAttribute(nsu, XMLSettingsHandler.XML_ATTRIBUTE_CLASS,
                    XMLSettingsHandler.XML_ATTRIBUTE_CLASS, nsu, mbeanInfo
                            .getClassName());
        }
        if (complexType.getParent() == null) {
            atts = new AttributesImpl();
        }
        handler.ignorableWhitespace(indentArray, 0, indent);
        handler.startElement(nsu, objectElement, objectElement, atts);
        for (Iterator it = complexType.getAttributeInfoIterator(settings); it
                .hasNext();) {
            ModuleAttributeInfo attribute = (ModuleAttributeInfo) it.next();
            if (!attribute.isTransient()) {
                parseAttribute(complexType, attribute, indent);
            }
        }
        handler.ignorableWhitespace(indentArray, 0, indent);
        handler.endElement(nsu, objectElement, objectElement);
    }

    private void parseAttribute(ComplexType complexType,
            ModuleAttributeInfo attribute, int indent) throws SAXException {
        Object value;
        try {
            value = complexType
                    .getLocalAttribute(settings, attribute.getName());
        } catch (AttributeNotFoundException e) {
            throw new SAXException(e);
        }
        if (orderFile || value != null) {
            // Write only overridden values unless this is the order file
            if (attribute.isComplexType()) {
                // Call method recursively for complex types
                parseComplexType((ComplexType) value, indent + indentAmount);
            } else {
                // Write element
                String elementName = SettingsHandler.getTypeName(attribute
                        .getType());
                AttributesImpl atts = new AttributesImpl();
                atts.addAttribute(nsu, XMLSettingsHandler.XML_ATTRIBUTE_NAME,
                        XMLSettingsHandler.XML_ATTRIBUTE_NAME, nsu, attribute
                                .getName());
                if (value == null) {
                    try {
                        value = complexType.getAttribute(attribute.getName());
                    } catch (Exception e) {
                        throw new SAXException(
                                "Internal error in settings subsystem", e);
                    }
                }
                if (value != null) {
                    handler.ignorableWhitespace(indentArray, 0, indent
                            + indentAmount);
                    handler.startElement(nsu, elementName, elementName, atts);
                    if (value instanceof ListType) {
                        parseListData(value, indent + indentAmount);
                        handler.ignorableWhitespace(indentArray, 0, indent
                                + indentAmount);
                    } else {
                        char valueArray[] = value.toString().toCharArray();
                        handler.characters(valueArray, 0, valueArray.length);
                    }
                    handler.endElement(nsu, elementName, elementName);
                }
            }
        }
    }

    /** Create SAX events for the content of a {@link ListType}.
     *
     * @param value the ListType whose content we create SAX events for.
     * @param indent the indentation amount for prettyprinting XML.
     * @throws SAXException is thrown if an error occurs.
     */
    private void parseListData(Object value, int indent) throws SAXException {
        ListType list = (ListType) value;
        Iterator it = list.iterator();
        while (it.hasNext()) {
            Object element = it.next();
            String elementName =
                SettingsHandler.getTypeName(element.getClass().getName());
            writeSimpleElement(
                elementName,
                element.toString(),
                null,
                indent + indentAmount);
        }
    }

    /** Resolve the XML element name of a {@link ComplexType}.
     *
     * @param complexType the object to investigate.
     * @return the name of the XML element.
     */
    private String resolveElementName(ComplexType complexType) {
        String elementName;
        if (complexType instanceof ModuleType) {
            if (complexType.getParent() == null) {
                // Top level controller element
                elementName = XMLSettingsHandler.XML_ELEMENT_CONTROLLER;
            } else if (
                !orderFile
                    && complexType.globalSettings().getModule(
                        complexType.getName())
                        != null) {
                // This is not the order file and we are referencing an object
                elementName = XMLSettingsHandler.XML_ELEMENT_OBJECT;
            } else {
                // The object is not referenced before
                elementName = XMLSettingsHandler.XML_ELEMENT_NEW_OBJECT;
            }
        } else {
            // It's a map
            elementName =
                SettingsHandler.getTypeName(complexType.getClass().getName());
        }
        return elementName;
    }

    /** Create SAX events for a simple element.
     *
     * Creates all the SAX events needed for prettyprinting an XML element
     * with a simple value and possible attributes.
     *
     * @param elementName the name of the XML element.
     * @param value the value to pu inside the XML element.
     * @param atts the attributes for the XML element.
     * @param indent the indentation amount for prettyprinting XML.
     * @throws SAXException is thrown if an error occurs.
     */
    private void writeSimpleElement(
        String elementName,
        String value,
        Attributes atts,
        int indent)
        throws SAXException {
        if (atts == null) {
            atts = new AttributesImpl();
        }
        // make sure that the value is never null
        value = value == null ? "" : value;
        handler.ignorableWhitespace(indentArray, 0, indent);
        handler.startElement(nsu, elementName, elementName, atts);
        handler.characters(value.toCharArray(), 0, value.length());
        handler.endElement(nsu, elementName, elementName);
    }

    /* (non-Javadoc)
     * @see org.xml.sax.XMLReader#parse(java.lang.String)
     */
    public void parse(String systemId) throws IOException, SAXException {
        // Do nothing. Just for conformance to the XMLReader API.
    }

    /* (non-Javadoc)
     * @see javax.xml.transform.sax.SAXSource#getXMLReader()
     */
    public XMLReader getXMLReader() {
        return this;
    }

    /* (non-Javadoc)
     * @see javax.xml.transform.sax.SAXSource#getInputSource()
     */
    public InputSource getInputSource() {
        return new InputSource();
    }
}