crawlsettingssaxhandler.java

这是个爬虫和lucece相结合最好了

            } else if (handlerStack.peek() instanceof RefinementHandler) {
                ((Refinement) stack.peek()).setDescription(value);
            } else {
                illegalElementError(name);
            }
        }
    }

    private class OrganizationHandler extends ElementHandler {

        public void endElement(String name) throws SAXException {
            if (handlerStack.peek() instanceof MetaHandler) {
                settings.setOrganization(value);
            } else if (handlerStack.peek() instanceof RefinementHandler) {
                ((Refinement) stack.peek()).setOrganization(value);
            } else {
                illegalElementError(name);
            }
        }
    }

    private class OperatorHandler extends ElementHandler {

        public void endElement(String name) throws SAXException {
            if (handlerStack.peek() instanceof MetaHandler) {
                settings.setOperator(value);
            } else if (handlerStack.peek() instanceof RefinementHandler) {
                ((Refinement) stack.peek()).setOperator(value);
            } else {
                illegalElementError(name);
            }
        }
    }

    private class AudienceHandler extends ElementHandler {

        public void endElement(String name) throws SAXException {
            if (handlerStack.peek() instanceof MetaHandler) {
                settings.setAudience(value);
            } else if (handlerStack.peek() instanceof RefinementHandler) {
                ((Refinement) stack.peek()).setAudience(value);
            } else {
                illegalElementError(name);
            }
        }
    }

    private class DateHandler extends ElementHandler {

        public void endElement(String name) throws SAXException {
            if (handlerStack.peek() instanceof MetaHandler) {
                try {
                    settings.setLastSavedTime(ArchiveUtils
                            .parse14DigitDate(value));
                } catch (ParseException e) {
                    throw new SAXException(e);
                }
            } else {
                illegalElementError(name);
            }
        }
    }

    // Refinement handlers
    private class RefinementListHandler extends ElementHandler {

        public void startElement(String name) throws SAXException {
            if (!(handlerStack.peek() instanceof RootHandler)) {
                illegalElementError(name);
            }
        }
    }

    private class RefinementHandler extends ElementHandler {
        public void startElement(String name, Attributes atts)
                throws SAXException {
            stack.push(new Refinement(settings, atts
                    .getValue(XMLSettingsHandler.XML_ELEMENT_REFERENCE)));
        }
    }

    private class ReferenceHandler extends ElementHandler {

        public void endElement(String name) throws SAXException {
            if (handlerStack.peek() instanceof RefinementHandler) {
                ((Refinement) stack.peek()).setReference(value);
            } else {
                illegalElementError(name);
            }
        }
    }

    private class LimitsHandler extends ElementHandler {
    }

    private class TimespanHandler extends ElementHandler {

        public void startElement(String name, Attributes atts)
                throws SAXException {
            if (stack.peek() instanceof Refinement) {
                String from = atts
                        .getValue(XMLSettingsHandler.XML_ATTRIBUTE_FROM);
                String to = atts.getValue(XMLSettingsHandler.XML_ATTRIBUTE_TO);
                try {
                    TimespanCriteria timespan = new TimespanCriteria(from, to);
                    ((Refinement) stack.peek()).addCriteria(timespan);
                } catch (ParseException e) {
                    throw new SAXException(e);
                }
            } else {
                illegalElementError(name);
            }
        }
    }

    private class PortnumberHandler extends ElementHandler {

        public void endElement(String name) throws SAXException {
            if (handlerStack.peek() instanceof LimitsHandler) {
                ((Refinement) stack.peek()).addCriteria(new PortnumberCriteria(value));
            } else {
                illegalElementError(name);
            }
        }
    }

    private class URIMatcherHandler extends ElementHandler {

        public void endElement(String name) throws SAXException {
            if (handlerStack.peek() instanceof LimitsHandler) {
                ((Refinement) stack.peek()).addCriteria(new RegularExpressionCriteria(value));
            } else {
                illegalElementError(name);
            }
        }
    }


    // Handlers for objects and attributes
    private class ModuleHandler extends ElementHandler {

        public void startElement(String name, Attributes atts)
                throws SAXException {
            ModuleType module;
            if (name.equals(XMLSettingsHandler.XML_ELEMENT_CONTROLLER)) {
                module = settingsHandler.getOrder();
            } else {
                module = settingsHandler.getSettingsObject(null).getModule(
                        atts.getValue(XMLSettingsHandler.XML_ATTRIBUTE_NAME));
            }
            stack.push(module);
        }

        public void endElement(String name) throws SAXException {
            stack.pop();
        }
    }

    private class NewModuleHandler extends ElementHandler {

        public void startElement(String name, Attributes atts)
                throws SAXException {
            ComplexType parentModule = (ComplexType) stack.peek();
            String moduleName = atts
                    .getValue(XMLSettingsHandler.XML_ATTRIBUTE_NAME);
            String moduleClass = atts
                    .getValue(XMLSettingsHandler.XML_ATTRIBUTE_CLASS);
            try {
                ModuleType module = SettingsHandler
                        .instantiateModuleTypeFromClassName(moduleName,
                                moduleClass);
                try {
                    parentModule.setAttribute(settings, module);
                } catch (AttributeNotFoundException e) {
                    // Attribute was not found, but the complex type might
                    // be a MapType and then we are allowed to add new
                    // elements.
                    try {
                        parentModule.addElement(settings, module);
                    } catch (IllegalStateException ise) {
                        // An attribute in the settings file is not in the
                        // ComplexType's definition, log and skip.
                        logger.log(Level.WARNING,"Module '" + moduleName + "' in '"
                                + locator.getSystemId() + "', line: "
                                + locator.getLineNumber() + ", column: "
                                + locator.getColumnNumber()
                                + " is not defined in '"
                                + parentModule.getName() + "'.");
                        throw new SAXException(new AttributeNotFoundException(
                                ise.getMessage()));
                    }
                }
                stack.push(module);
            } catch (InvocationTargetException e) {
                logger.log(Level.WARNING,"Couldn't instantiate " + moduleName
                        + ", from class: " + moduleClass + "' in '"
                        + locator.getSystemId() + "', line: "
                        + locator.getLineNumber() + ", column: "
                        + locator.getColumnNumber(), e);
                throw new SAXException(e);
            } catch (InvalidAttributeValueException e) {
                throw new SAXException(e);
            }
        }

        public void endElement(String name) throws SAXException {
            stack.pop();
        }
    }

    private class MapHandler extends ElementHandler {

        public void startElement(String name, Attributes atts)
                throws SAXException {
            String mapName = atts
                    .getValue(XMLSettingsHandler.XML_ATTRIBUTE_NAME);
            ComplexType parentModule = (ComplexType) stack.peek();
            try {
                stack.push(parentModule.getAttribute(settings, mapName));
            } catch (AttributeNotFoundException e) {
                throw new SAXException(e);
            }
        }

        public void endElement(String name) throws SAXException {
            stack.pop();
        }
    }

    private class SimpleElementHandler extends ElementHandler {

        public void startElement(String name, Attributes atts)
                throws SAXException {
            stack.push(atts.getValue(XMLSettingsHandler.XML_ATTRIBUTE_NAME));
        }

        public void endElement(String name) throws SAXException {
            String elementName = (String) stack.pop();
            Object container = stack.peek();
            if (container instanceof ComplexType) {
                try {
                    try {
                        ((ComplexType) container).setAttribute(settings,
                                new Attribute(elementName, value));
                    } catch (AttributeNotFoundException e) {
                        // Attribute was not found, but the complex type might
                        // be a MapType and then we are allowed to add new
                        // elements.
                        try {
                            ((ComplexType) container).addElement(settings,
                                    new SimpleType(elementName, "", value));
                        } catch (IllegalStateException ise) {
                            logger.warning("Unknown attribute '" + elementName
                                    + "' in '" + locator.getSystemId()
                                    + "', line: " + locator.getLineNumber()
                                    + ", column: " + locator.getColumnNumber());
                        }
                    }
                } catch (InvalidAttributeValueException e) {
                    try {
                        logger.warning("Illegal value '"
                                + value
                                + "' for attribute '"
                                + elementName
                                + "' in '"
                                + locator.getSystemId()
                                + "', line: "
                                + locator.getLineNumber()
                                + ", column: "
                                + locator.getColumnNumber()
                                + ", Value reset to default value: "
                                + ((ComplexType) container).getAttribute(
                                        settings, elementName));
                    } catch (AttributeNotFoundException e1) {
                        throw new SAXException(e1);
                    }
                }
            } else {
                if (container == null) {
                	// We can get here if an override is referring to a global
                    // filter since removed.  Log it as severe; operator will
                    // probably want to know of all overrides with references
                    // to a global filter since removed.
                    logger.severe("Empty container (Was a referenced parent" +
                        " filter removed?).  Element details: elementName " +
                        elementName + ", name " + name);
                } else {
                	((ListType) container).add(value);
                }
            }
        }
    }

    private class ListHandler extends ElementHandler {

        public void startElement(String name, Attributes atts)
                throws SAXException {
            String listName = atts
                    .getValue(XMLSettingsHandler.XML_ATTRIBUTE_NAME);
            ComplexType parentModule = (ComplexType) stack.peek();
            ListType list;
            try {
                list = (ListType) parentModule.getAttribute(settings, listName);
            } catch (AttributeNotFoundException e) {
                throw new SAXException(e);
            }
            list.clear();
            stack.push(list);
        }

        public void endElement(String name) throws SAXException {
            stack.pop();
        }
    }

    /*
     * (non-Javadoc)
     *
     * @see org.archive.crawler.settings.ValueErrorHandler#handleValueError(org.archive.crawler.settings.Constraint.FailedCheck)
     */
    public void handleValueError(FailedCheck error) {
        logger.warning(error.getMessage() + "\n Attribute: '"
                + error.getOwner().getName() + ":"
                + error.getDefinition().getName() + "'\n Value:     '" + value
                + "'\n File:      '" + locator.getSystemId() + "', line: "
                + locator.getLineNumber() + ", column: "
                + locator.getColumnNumber());
    }
}