看了几个网上实现的纯真IP库的解析程序，大都是通过随机文件访问类RandomAccessFile或者内存映射MappedByteBuffer的方式来读取库文件。但是这种方式在高并发的环境下会出错，比如： http://www.iteye.com/topic/340548 这篇例子中的IPSeeker类getIPLocation(long offset)方法中
loc.setCountry ( readString(ipFile.getFilePointer() - 1)); loc.setArea( readArea(ipFile.getFilePointer()));
setCountry方法执行过后会修改文件的位置指针，然后setArea方法会依赖于上一个方法结束后的文件位置指针。由于ipFile是公有变量，所以在高并发的环境下会导致指针错乱。
我们的业务需求只是输入IP查询信息，因此将上面的例子进行了修改，去掉了无关的方法。然后通过InputStream字节流一次性读取IP库文件将其保存在byte数组中，操作数组来读取IP信息，解决并发问题。
IPSeeker
import java.io.IOException;
import java.io.InputStream;
import java.util.concurrent.ConcurrentHashMap;
import org.apache.log4j.Logger;
public class IPSeeker {
Logger logger = Logger.getLogger(IPSeeker.class);
// 一些固定常量，比如记录长度等等
private static final int IP_RECORD_LENGTH = 7;
private static final byte REDIRECT_MODE_1 = 0x01;
private static final byte REDIRECT_MODE_2 = 0x02;
private static final String FILENAME = "qqwry.dat";
// 用来做为cache，查询一个ip时首先查看cache，以减少不必要的重复查找
private ConcurrentHashMap<String, IPLocation> ipCache;
// 起始地区的开始和结束的绝对偏移
private long ipBegin, ipEnd;
private static IPSeeker instance = null;
private static IPByteArray byteArray = null;
/**
* 构造函数，传入值为InputStream
* @param inputStream
* @return
*/
public static IPSeeker getInstance(InputStream inputStream){
if( null==instance ){
synchronized(IPSeeker.class){
if( null==instance ){
byteArray = new IPByteArray(IPLocationUtil.getByteArrayFromInputStream(inputStream));
instance = new IPSeeker();
}
}
}
return instance;
}
/**
* 构造函数，传入文件默认Classpath下qqwry.dat
* @param fileName
* @return
*/
public static IPSeeker getInstance(String fileName){
if( null==instance ){
synchronized(IPSeeker.class){
if( null==instance ){
byteArray = new IPByteArray(IPLocationUtil.getByteArrayFromClasspathFile(fileName));
instance = new IPSeeker();
}
}
}
return instance;
}
/**
* 构造函数，传入值为Classpath下的文件名称
* @param fileName
* @return
*/
public static IPSeeker getInstance(){
if( null==instance ){
synchronized(IPSeeker.class){
if( null==instance ){
byteArray = new IPByteArray(IPLocationUtil.getByteArrayFromClasspathFile(FILENAME));
instance = new IPSeeker();
}
}
}
return instance;
}
private IPSeeker() {
ipCache = new ConcurrentHashMap<String, IPLocation>();
ipBegin = readLong4(0);
ipEnd = readLong4(4);
}
/**
* 查询IP信息，返回IPLocation对象
* @param ip
* @return
*/
public IPLocation getIPLocation(String ip) {
byte[] ipByte = IPLocationUtil.getIpByteArrayFromString(ip);
if (ipCache.containsKey(ip)) {
IPLocation ipLoc = ipCache.get(ip);
return ipLoc;
} else {
IPLocation ipLoc = getIPLocation(ipByte);
ipCache.put(ip, ipLoc.getCopy());
return ipLoc;
}
}
/**
* 查询IP信息，返回country+" "+area字符串
* @param ip
* @return
*/
public String getIPAdress(String ip){
IPLocation ipLocation = getIPLocation(ip);
String country = ipLocation.getCountry();
if(country != null && country.length() > 1){
String temp = country.substring(0,2);
if(IPLocationUtil.provinceMap.containsKey(temp)){
country = IPLocationUtil.provinceMap.get(temp).toString();
}
}
String address = country + " " + ipLocation.getArea();
return address.trim();
}
/**
* 根据ip搜索ip信息文件，得到IPLocation结构，所搜索的ip参数从类成员ip中得到
*
* @param ip 要查询的IP
* @return IPLocation结构
*/
private IPLocation getIPLocation(byte[] ip) {
IPLocation info = null;
long offset = locateIP(ip);
if (offset != -1)
info = getIPLocation(offset);
if (info == null) {
info = new IPLocation();
info.setCountry(Message.unknown_country);
info.setArea(Message.unknown_area);
}
return info;
}
/**
* 从offset位置读取4个字节为一个long，因为java为big-endian格式，所以没办法 用了这么一个函数来做转换
*
* @param offset
* @return 读取的long值，返回-1表示读取文件失败
*/
private long readLong4(long offset) {
byte[] byte4 = new byte[4];
long ret = 0;
try {
byteArray.read((int)offset, byte4);
ret |= (byte4[0] & 0xFF);
ret |= ((byte4[1] << 8) & 0xFF00);
ret |= ((byte4[2] << 16) & 0xFF0000);
ret |= ((byte4[3] << 24) & 0xFF000000);
return ret;
} catch (Exception e) {
e.printStackTrace();
return -1;
}
}
/**
* 从offset位置读取3个字节为一个long，因为java为big-endian格式，所以没办法 用了这么一个函数来做转换
*
* @param offset 整数的起始偏移
* @return 读取的long值，返回-1表示读取文件失败
*/
private long readLong3(long offset) {
byte[] b3 = new byte[3];
long ret = 0;
try {
byteArray.read((int)offset, b3);
ret |= (b3[0] & 0xFF);
ret |= ((b3[1] << 8) & 0xFF00);
ret |= ((b3[2] << 16) & 0xFF0000);
return ret;
} catch (Exception e) {
return -1;
}
}
/**
* 从offset位置读取四个字节的ip地址放入ip数组中，读取后的ip为big-endian格式，但是 文件中是little-endian形式，将会进行转换
*
* @param offset
* @param ip
*/
private void readIP(long offset, byte[] ip) {
try {
byteArray.read((int)offset,ip);
byte temp = ip[0];
ip[0] = ip[3];
ip[3] = temp;
temp = ip[1];
ip[1] = ip[2];
ip[2] = temp;
} catch (Exception e) {
logger.error(e.toString());
}
}
/**
* 把类成员ip和beginIp比较，注意这个beginIp是big-endian的
*
* @param ip 要查询的IP
* @param beginIp 和被查询IP相比较的IP
* @return 相等返回0，ip大于beginIp则返回1，小于返回-1。
*/
private int compareIP(byte[] ip, byte[] beginIp) {
for (int i = 0; i < 4; i++) {
int r = compareByte(ip[i], beginIp[i]);
if (r != 0)
return r;
}
return 0;
}
/**
* 把两个byte当作无符号数进行比较
*
* @param b1
* @param b2
* @return 若b1大于b2则返回1，相等返回0，小于返回-1
*/
private int compareByte(byte b1, byte b2) {
if ((b1 & 0xFF) > (b2 & 0xFF)) // 比较是否大于
return 1;
else if ((b1 ^ b2) == 0)// 判断是否相等
return 0;
else
return -1;
}
/**
* 这个方法将根据ip的内容，定位到包含这个ip国家地区的记录处，返回一个绝对偏移 方法使用二分法查找。
*
* @param ip 要查询的IP
* @return 如果找到了，返回结束IP的偏移，如果没有找到，返回-1
*/
private long locateIP(byte[] ip) {
byte[] b4 = new byte[4];
long m = 0;
int r;
// 比较第一个ip项
readIP(ipBegin, b4);
r = compareIP(ip, b4);
if (r == 0)
return ipBegin;
else if (r < 0)
return -1;
// 开始二分搜索
for (long i = ipBegin, j = ipEnd; i < j;) {
m = getMiddleOffset(i, j);
readIP(m, b4);
r = compareIP(ip, b4);
if (r > 0)
i = m;
else if (r < 0) {
if (m == j) {
j -= IP_RECORD_LENGTH;
m = j;
} else
j = m;
} else
return readLong3(m + 4);
}
// 如果循环结束了，那么i和j必定是相等的，这个记录为最可能的记录，但是并非
// 肯定就是，还要检查一下，如果是，就返回结束地址区的绝对偏移
m = readLong3(m + 4);
readIP(m, b4);
r = compareIP(ip, b4);
if (r <= 0)
return m;
else
return -1;
}
/**
* 得到begin偏移和end偏移中间位置记录的偏移
*
* @param begin
* @param end
* @return
*/
private long getMiddleOffset(long begin, long end) {
long records = (end - begin) / IP_RECORD_LENGTH;
records >>= 1;
if (records == 0)
records = 1;
return begin + records * IP_RECORD_LENGTH;
}
/**
* 给定一个ip国家地区记录的偏移，返回一个IPLocation结构
*
* @param offset 国家记录的起始偏移
* @return IPLocation对象
*/
private IPLocation getIPLocation(long offset) {
try {
IPLocation loc = new IPLocation();
// 跳过4字节ip
long position = offset+4;
// 读取第一个字节判断是否标志字节
byte b = byteArray.read((int)position);
position++;
if (b == REDIRECT_MODE_1) {
// 读取国家偏移
long countryOffset = readLong3(position);
// 跳转至偏移处
position = countryOffset;
// 再检查一次标志字节，因为这个时候这个地方仍然可能是个重定向
b = byteArray.read((int)position);
position++;
if (b == REDIRECT_MODE_2) {
readCountry(loc,readLong3(position));
readArea(loc,countryOffset+4);
} else {
long afterReadCountry = readCountry(loc,countryOffset);
readArea(loc,afterReadCountry);
}
} else if (b == REDIRECT_MODE_2) {
readCountry(loc,readLong3(position));
readArea(loc,offset+8);
} else {
long afterReadCountry = readCountry(loc, position-1);
readArea(loc, afterReadCountry);
}
return loc;
} catch (IOException e) {
return null;
}
}
private long readCountry(IPLocation loc, long offset) throws IOException {
return readString( loc, (int)offset, 1 );
}
private void readArea(IPLocation loc, long offset) throws IOException {
byte b = byteArray.read((int)offset);
if (b == REDIRECT_MODE_1 || b == REDIRECT_MODE_2) {
long areaOffset = readLong3(offset + 1);
if (areaOffset == 0){
loc.setArea(Message.unknown_area);
}else {
readString( loc, (int)areaOffset, 2 );
}
} else {
readString( loc, (int)offset, 2 );
}
}
/**
* 从position偏移处读取一个以0结束的字符串
* @param loc
* @param position 字符串起始偏移
* @param type
* @return
*/
private int readString(IPLocation loc, int position, int type){
int n=0;
byte[] temp = new byte[100];
while( true ){
if(byteArray.read(position) == 0){
break;
} else {
temp[n] = byteArray.read(position);
position++;
n++;
}
if (n >= temp.length) {
byte[] tmp = new byte[n + 100];
System.arraycopy(temp, 0, tmp, 0, n);
temp = tmp;
}
}
if (n != 0){
if( type==1 )
loc.setCountry(IPLocationUtil.getString(temp, 0, n, "GBK"));
else
loc.setArea(IPLocationUtil.getString(temp, 0, n, "GBK"));
}
return position+1;
}
}
IPByteArray
扫描二维码关注公众号，回复：
823399 查看本文章
import org.apache.log4j.Logger;
public class IPByteArray {
Logger logger = Logger.getLogger(IPByteArray.class);
private byte[] byteArray;
public IPByteArray( byte[] byteArray ){
this.byteArray = byteArray;
}
public void read( int position, byte[] bytes ){
int p = position;
for( int i=0; i<bytes.length; i++ ){
bytes[i] = read(p);
p++;
}
}
public byte read(int position){
return byteArray[position];
}
}
IPLocationUtil
import java.io.IOException;
import java.io.InputStream;
import java.io.UnsupportedEncodingException;
import java.util.HashMap;
import java.util.Map;
import java.util.StringTokenizer;
import org.apache.log4j.Logger;
public class IPLocationUtil {
public static Logger logger = Logger.getLogger(IPLocationUtil.class);
/**
* 将Classpath下文件转为Byte数组
* @param fileName
* @return
*/
public static byte[] getByteArrayFromClasspathFile(String fileName){
InputStream inputStream = ClassLoaderUtil.getResourceAsStream(fileName, IPLocationUtil.class);
byte[] array = getByteArrayFromInputStream(inputStream);
try {
inputStream.close();
} catch (IOException e) {
logger.error("关闭InputStream错误", e);
}
return array;
}
/**
* 将InputStream转换为Byte数组
* @param inputStream
* @return
*/
public static byte[] getByteArrayFromInputStream(InputStream inputStream){
byte[] array = null;
try {
array = new byte[inputStream.available()];
inputStream.read(array, 0, array.length);
} catch (IOException e) {
logger.error("将InputStream转换为Byte数组错误", e);
}
return array;
}
/**
* 从ip的字符串形式得到字节数组形式
* @param ip 字符串形式的ip
* @return 字节数组形式的ip
*/
public static byte[] getIpByteArrayFromString(String ip) {
byte[] ret = new byte[4];
StringTokenizer st = new StringTokenizer(ip, ".");
try {
ret[0] = (byte)(Integer.parseInt(st.nextToken()) & 0xFF);
ret[1] = (byte)(Integer.parseInt(st.nextToken()) & 0xFF);
ret[2] = (byte)(Integer.parseInt(st.nextToken()) & 0xFF);
ret[3] = (byte)(Integer.parseInt(st.nextToken()) & 0xFF);
} catch (Exception e) {
logger.error("从ip的字符串形式得到字节数组形式报错"+e.toString());
}
return ret;
}
/**
* 根据某种编码方式将字节数组转换成字符串
* @param b 字节数组
* @param offset 要转换的起始位置
* @param len 要转换的长度
* @param encoding 编码方式
* @return 如果encoding不支持，返回一个缺省编码的字符串
*/
public static String getString(byte[] b, int offset, int len, String encoding) {
try {
return new String(b, offset, len, encoding);
} catch (UnsupportedEncodingException e) {
return new String(b, offset, len);
}
}
}