Android实战技巧：用Pull方法解析XML文件

Pull解析方法给应用程序完全的控制文档该怎么样被解析。Android中对Pull方法提供了支持的API，主要是

org.xmlpull.v1.XmlPullParser;
org.xmlpull.v1.XmlPullParserFactory;

二个类，其中主要使用的是XmlPullParser，XmlPullParserFactory是一个工厂，用于构建XmlPullParser对象。
应用程序通过调用XmlPullParser.next()等方法来产生Event，然后再处理Event。可以看到它与Push方法的不同，Push方法是由Parser自己主动产生Event，回调给应用程序。而Pull方法是主动的调用Parser的方法才能产生事件。
假如XML中的语句是这样的："<author country="United States">James Elliott</author>"，author是TAG，country是ATTRIBUTE，"James Elliott"是TEXT。
要想解析文档先要构建一个XmlPullParser对象

        final XmlPullParserFactory factory = XmlPullParserFactory.newInstance();
        factory.setNamespaceAware(true);
        final XmlPullParser parser = factory.newPullParser();

Pull解析是一个遍历文档的过程，每次调用next()，nextTag(), nextToken()和nextText()都会向前推进文档，并使Parser停留在某些事件上面，但是不能倒退。
然后把文档设置给Parser

        parser.setInput(new StringReader("<author country=\"United States\">James Elliott</author>");

这时，文档刚被初始化，所以它应该位于文档的开始，事件应该是START_DOCUMENT，可以通过XmlPullParser.getEventType()来获取。然后调用next()会产生

• START_TAG，这个事件告诉应用程序一个标签已经开始了，调用getName()会返回"author";再next()会产生

• TEXT事件，调用getText()会返回"James Elliott"，再next()，会产生

• END_TAG，这个告诉你一个标签已经处理完了，再next()，会产生

• END_DOCUMENT，它告诉你整个文档已经处理完成了。

除了next()外，nextToken()也可以使用，只不过它会返回更加详细的事件，比如 COMMENT, CDSECT, DOCDECL, ENTITY等等非常详细的信息。如果程序得到比较底层的信息，可以用nextToken()来驱动并处理详细的事件。需要注意一点的是TEXT事件是有可能返回空白的White Spaces比如换行符或空格等。
另外有二个非常实用的方法nextTag()和nextText()
nextTag()--首先它会忽略White Spaces，如果可以确定下一个是START_TAG或END_TAG，就可以调用nextTag()直接跳过去。通常它有二个用处：当START_TAG时，如果能确定这个TAG含有子TAG，那么就可以调用nextTag()产生子标签的START_TAG事件；当END_TAG时，如果确定不是文档结尾，就可以调用nextTag()产生下一个标签的START_TAG。在这二种情况下如果用next()会有TEXT事件，但返回的是换行符或空白符。
nextText()--它只能在START_TAG时调用。当下一个元素是TEXT时，TEXT的内容会返回；当下一个元素是END_TAG时，也就是说这个标签的内容为空，那么空字串返回；这个方法返回后，Parser会停在END_TAG上。比如：

<author>James Elliott</author>
<author></author>
<author/>

当START_TAG时，调用nextText()，依次返回：
"James Elliott"
""(empty)
""(empty)
这个方法在处理没有子标签的标签时很有用。比如：

<title>What Is Hibernate</title>
<author>James Elliott</author>
<category>Web</category>

就可以用以下代码来处理：

        while (eventType != XmlPullParser.END_TAG) {
            switch (eventType) {
            case XmlPullParser.START_TAG:
                tag = parser.getName();
                final String content = parser.nextText();
                Log.e(TAG, tag + ": [" + content + "]");
                eventType = parser.nextTag();
                break;
            default:
                break;
            }
        }

这就要比用next()来处理方便多了，可读性也大大的加强了。
最后附上一个解析XML的实例Android程序

import java.io.IOException;
import java.io.InputStream;

import org.xmlpull.v1.XmlPullParser;
import org.xmlpull.v1.XmlPullParserException;
import org.xmlpull.v1.XmlPullParserFactory;

import android.util.Log;

public class RssPullParser extends RssParser {
    private final String TAG = FeedSettings.GLOBAL_TAG;
    
    private InputStream mInputStream;
    
    public RssPullParser(InputStream is) {
        mInputStream = is;
    }
    
    public void parse() throws ReaderBaseException, XmlPullParserException, IOException {
        if (mInputStream == null) {
            throw new ReaderBaseException("no input source, did you initialize this class correctly?");
        }
        final XmlPullParserFactory factory = XmlPullParserFactory.newInstance();
        factory.setNamespaceAware(true);
        final XmlPullParser parser = factory.newPullParser();
        
        parser.setInput(mInputStream);
        int eventType = parser.getEventType();
        if (eventType != XmlPullParser.START_DOCUMENT) {
            throw new ReaderBaseException("Not starting with 'start_document'");
        }
        eventType = parseRss(parser);
        if (eventType != XmlPullParser.END_DOCUMENT) {
            throw new ReaderBaseException("not ending with 'end_document', do you finish parsing?");
        }
        if (mInputStream != null) {
            mInputStream.close();
        } else {
            Log.e(TAG, "inputstream is null, XmlPullParser closed it??");
        }
    }
    
    /**
     * Parsing the Xml document. Current type must be Start_Document.
     * After calling this, Parser is positioned at END_DOCUMENT.
     * @param parser
     * @return event end_document
     * @throws XmlPullParserException
     * @throws ReaderBaseException
     * @throws IOException
     */
    private int parseRss(XmlPullParser parser) throws XmlPullParserException, ReaderBaseException, IOException {
        int eventType = parser.getEventType();
        if (eventType != XmlPullParser.START_DOCUMENT) {
            throw new ReaderBaseException("not starting with 'start_document', is this a new document?");
        }
        Log.e(TAG, "starting document, are you aware of that!");
        eventType = parser.next();
        while (eventType != XmlPullParser.END_DOCUMENT) {
            switch (eventType) {
            case XmlPullParser.START_TAG: {
                Log.e(TAG, "start tag: '" + parser.getName() + "'");
                final String tagName = parser.getName();
                if (tagName.equals(RssFeed.TAG_RSS)) {
                    Log.e(TAG, "starting an RSS feed <<");
                    final int attrSize = parser.getAttributeCount();
                    for (int i = 0; i < attrSize; i++) {
                        Log.e(TAG, "attr '" + parser.getAttributeName(i) + "=" + parser.getAttributeValue(i) + "'");
                    }
                } else if (tagName.equals(RssFeed.TAG_CHANNEL)) {
                    Log.e(TAG, "\tstarting an Channel <<");
                    parseChannel(parser);
                }
                break;
            }
            case XmlPullParser.END_TAG: {
                Log.e(TAG, "end tag: '" + parser.getName() + "'");
                final String tagName = parser.getName();
                if (tagName.equals(RssFeed.TAG_RSS)) {
                    Log.e(TAG, ">> edning an RSS feed");
                } else if (tagName.equals(RssFeed.TAG_CHANNEL)) {
                    Log.e(TAG, "\t>> ending an Channel");     
                }
                break;
            }
            default:
                break;
            }
            eventType = parser.next();
        }
        Log.e(TAG, "end of document, it is over");
        return parser.getEventType();
    }
    
    /**
     * Parse a channel. MUST be start tag of an channel, otherwise exception thrown.
     * Param XmlPullParser
     * After calling this function, parser is positioned at END_TAG of Channel.
     * return end tag of a channel
     * @throws XmlPullParserException
     * @throws ReaderBaseException
     * @throws IOException
     */
    private int parseChannel(XmlPullParser parser) throws XmlPullParserException, ReaderBaseException, IOException {
        int eventType = parser.getEventType();
        String tagName = parser.getName();
        if (eventType != XmlPullParser.START_TAG || !RssFeed.TAG_CHANNEL.equals(tagName)) {
            throw new ReaderBaseException("not start with 'start tag', is this a start of a channel?");
        }
        Log.e(TAG, "\tstarting " + tagName);
        eventType = parser.nextTag();
        while (eventType != XmlPullParser.END_TAG) {
            switch (eventType) {
            case XmlPullParser.START_TAG: {
                final String tag = parser.getName();
                if (tag.equals(RssFeed.TAG_IMAGE)) {
                    parseImage(parser);
                } else if (tag.equals(RssFeed.TAG_ITEM)) {
                    parseItem(parser);
                } else {
                    final String content = parser.nextText();
                    Log.e(TAG, tag + ": [" + content + "]");
                }
                // now it SHOULD be at END_TAG, ensure it
                if (parser.getEventType() != XmlPullParser.END_TAG) {
                    throw new ReaderBaseException("not ending with 'end tag', did you finish parsing sub item?");
                }
                eventType = parser.nextTag();
                break;
            }
            default:
                break;
            }
        }
        Log.e(TAG, "\tending " + parser.getName());
        return parser.getEventType();
    }
    
    /**
     * Parse image in a channel.
     * Precondition: position must be at START_TAG and tag MUST be 'image'
     * Postcondition: position is END_TAG of '/image'
     * @throws IOException
     * @throws XmlPullParserException
     * @throws ReaderBaseException
     */
    private int parseImage(XmlPullParser parser) throws XmlPullParserException, IOException, ReaderBaseException {
        int eventType = parser.getEventType();
        String tag = parser.getName();
        if (eventType != XmlPullParser.START_TAG || !RssFeed.TAG_IMAGE.equals(tag)) {
            throw new ReaderBaseException("not start with 'start tag', is this a start of an image?");
        }
        Log.e(TAG, "\t\tstarting image " + tag);
        eventType = parser.nextTag();
        while (eventType != XmlPullParser.END_TAG) {
            switch (eventType) {
            case XmlPullParser.START_TAG:
                tag = parser.getName();
                Log.e(TAG, tag + ": [" + parser.nextText() + "]");
                // now it SHOULD be at END_TAG, ensure it
                if (parser.getEventType() != XmlPullParser.END_TAG) {
                    throw new ReaderBaseException("not ending with 'end tag', did you finish parsing sub item?");
                }
                eventType = parser.nextTag();
                break;
            default:
                break;
            }
        }
        Log.e(TAG, "\t\tending image " + parser.getName());
        return parser.getEventType();
    }
    
    /**
     * Parse an item in a channel.
     * Precondition: position must be at START_TAG and tag MUST be 'item'
     * Postcondition: position is END_TAG of '/item'
     * @throws IOException
     * @throws XmlPullParserException
     * @throws ReaderBaseException
     */
    private int parseItem(XmlPullParser parser) throws XmlPullParserException, IOException, ReaderBaseException {
        int eventType = parser.getEventType();
        String tag = parser.getName();
        if (eventType != XmlPullParser.START_TAG || !RssFeed.TAG_ITEM.equals(tag)) {
            throw new ReaderBaseException("not start with 'start tag', is this a start of an item?");
        }
        Log.e(TAG, "\t\tstarting " + tag);
        eventType = parser.nextTag();
        while (eventType != XmlPullParser.END_TAG) {
            switch (eventType) {
            case XmlPullParser.START_TAG:
                tag = parser.getName();
                final String content = parser.nextText();
                Log.e(TAG, tag + ": [" + content + "]");

                // now it SHOULD be at END_TAG, ensure it
                if (parser.getEventType() != XmlPullParser.END_TAG) {
                    throw new ReaderBaseException("not ending with 'end tag', did you finish parsing sub item?");
                }
                eventType = parser.nextTag();
                break;
            default:
                break;
            }
        }
        Log.e(TAG, "\t\tending " + parser.getName());
        return parser.getEventType();
    }
}